waspinator / pycococreator Goto Github PK

View Code? Open in Web Editor NEW

770.0 19.0 179.0 180 KB

Helper functions to create COCO datasets

License: Apache License 2.0

Python 100.00%

computer-vision machine-learning dataset coco

pycococreator's Introduction

pycococreator

Please site using:

pycococreator is a set of tools to help create COCO datasets. It includes functions to generate annotations in uncompressed RLE ("crowd") and polygons in the format COCO requires.

Install

pip install git+git://github.com/waspinator/[email protected]

If you need to install pycocotools for python 3, try the following:

sudo apt-get install python3-dev
pip install cython
pip install git+git://github.com/waspinator/[email protected]

pycococreator's People

Contributors

Stargazers

Watchers

Forkers

zumbalamambo shubhampachori12110095 kekedan clscy arangrej yuchunliu andytung2019 anjanakn kaidi88 chengyangdu levelsethu hannarud lanshanikilven fan5ga ted8201 visionnoob xm1112 umicoo russel-sht thiagodangelo miaowu99 winwinjjiang dmaugis henny1011 shadowkun meynmd xu-jt jiazewang asyncbridge fegonda shanlans faninafanina zm66260 blackangel1111 kukhyunlim exploding-gradients rodney-wang rmartella husband-of-reimu salvador-blanco aminsung kinghao93 jianxiongcai joskaaaa ella2le mejihero kobewangsky aliborji william-zhan svats2k xuyunqiu janus1999 perif junjiwon1031 bybysker hongwei94 wanghuaichen xiong224 fred-fan nonemountain absaravanan pkusnail juzisedefeimao zhenguoyuan calvine finesure2017 seungback kp-forks i-alpha baoyufuyou nuaasxr yyuzhong mygit007hub solauky hanyeliu zouxiaodong shalevy1 liangxu123 zgsxwsdxg ssjatmhmy chaoso smidm blyucs isrkhou tobias-fyi yaoliuoa chadng m-kashani itamargru kdh4672 liu199604 timofurrer chrissem johnpapa97 eileen2014 jiheeseon jinlinyi kevinn1999 tata24 shuhanstack

pycococreator's Issues

How to run?

Could you please provide a little bit more clearer description of how to run your code? I installed the package and a little bit unsure of what to do next. Trying to run pycococreatortools.py, but it says that numpy is not installed (but it installed).

Getting error when trying to install in Kaggle

After running !pip install git+git://github.com/waspinator/[email protected]

getting errror

Collecting git+git://github.com/waspinator/[email protected]
  Cloning git://github.com/waspinator/pycococreator.git (to revision 0.2.0) to /tmp/pip-req-build-m589ikd6
  Running command git clone --filter=blob:none --quiet git://github.com/waspinator/pycococreator.git /tmp/pip-req-build-m589ikd6
  fatal: remote error:
    The unauthenticated git protocol on port 9418 is no longer supported.
  Please see https://github.blog/2021-09-01-improving-git-protocol-security-github/ for more information.
  error: subprocess-exited-with-error
  
  × git clone --filter=blob:none --quiet git://github.com/waspinator/pycococreator.git /tmp/pip-req-build-m589ikd6 did not run successfully.
  │ exit code: 128
  ╰─> See above for output.
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error

× git clone --filter=blob:none --quiet git://github.com/waspinator/pycococreator.git /tmp/pip-req-build-m589ikd6 did not run successfully.
│ exit code: 128
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.

Support multiple annotations in the same image of the same class

I'd be very useful to get multiple annotations from a single binary mask whereas disconnected shapes are separate annotations instead of one.

Is that something which could make it into this package?

Preparing Custom Dataset - "NameError: name 'fnmatch' is not defined"

I have a dataset folder which contains two folders named images, annotations(../train/images, annotations). Each of these folders contains images. For example:

In images:

In annotations:

I have seen this blog. I modified the code and tried to run in colab for generating dataset. Then I got this error: NameError: name 'fnmatch' is not defined

How can I solve the issue? Can anyone guide me through the procedure for creating a dataset to apply mask rcnn?
@waspinator @hannarud

I got the original image as output

When visualizing with own images: ValueError: cannot reshape array of size 5 into shape (2,2)

Hey, I recreated the steps with my own image as attached. Adapting your shapes_to_coco.py script by changing class names it worked. However, in the last step of visualizing them with your jupyter notebook (there is an error, the annotation_file is not in the annotations dir but in the train dir) I get the following error: ValueError: cannot reshape array of size 5 into shape (2,2). Screenshot is attached too.

Maybe we can discuss this in detail via mail to speed up fixing it: [email protected]

Failed to transform dataset with figures as filename

Just want to raise your attention, for the following test case,

./train/annotations/139284-0989-1254.txt

failed to transform to annotation in json format. I assume that your framework should support this.

Can you leave me some instructions to counter this?

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.

(venv) root@cc3180440f4b:/opendr/src/opendr/perception/panoptic_segmentation/efficient_ps/algorithm/EfficientPS# python tools/convert_cityscapes.py ./data_2/ ./data/cityscapes/
Loading Cityscapes from ./data_2/
Converting train ...
0%| | 0/2975 [00:00<?, ?it/s]
multiprocessing.pool.RemoteTraceback:
"""
Traceback (most recent call last):
File "/usr/lib/python3.8/multiprocessing/pool.py", line 125, in worker
result = (True, func(*args, **kwds))
File "/usr/lib/python3.8/multiprocessing/pool.py", line 48, in mapstar
return list(map(*args))
File "tools/convert_cityscapes.py", line 163, in call
coco_ann_i = pct.create_annotation_info(
File "/opendr/venv/lib/python3.8/site-packages/pycococreatortools/pycococreatortools.py", line 99, in create_annotation_info
segmentation = binary_mask_to_polygon(binary_mask, tolerance)
File "/opendr/venv/lib/python3.8/site-packages/pycococreatortools/pycococreatortools.py", line 48, in binary_mask_to_polygon
contours = np.subtract(contours, 1)
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.

I am trying to convert my Citscapes dataset to COCO style using pycococreator with following subfolder.

gtFine
leftImg8bit

Inside gtFine->train-><city_name>->following items
1.aachen_000057_000019_gtFine_color.png
aachen_000057_000019_gtFine_instanceIds.png
aachen_000057_000019_gtFine_labelIds.png
aachen_000057_000019_gtFine_polygons.json

How can I solve the above ValueError issue?

TypeError: Argument 'bb' has incorrect type (expected numpy.ndarray, got list)

I'm getting this while trying to train the matterport mask rcnn model on my own data created with this tool. Anybody know what this is caused by? Another person opened this issue here: cocodataset/cocoapi#139 but closed it without saying how to fix.

IndexError: list index out of range

Full error:

Traceback (most recent call last):
  File "segment.py", line 108, in <module>
    class_id = [x['id'] for x in CATEGORIES if x['name'] in annotation_filename][0]
IndexError: list index out of range

Here's the code:

INFO = {
    "description": "Fashion Dataset",
    "url": "https://github.com/waspinator/pycococreator",
    "version": "0.1.0",
    "year": 2020,
    "contributor": "Abu Noman Md. Sakib",
    "date_created": datetime.datetime.utcnow().isoformat(' ')
}

LICENSES = [
    {
        "id": 1,
        "name": "GB",
        "url": "GB"
    }
]

CATEGORIES = [
    {
        'id': 1,
        'name': 'mask',
        'supercategory': 'fashion',
    }
]

coco_output = {
    "info": INFO,
    "licenses": LICENSES,
    "categories": CATEGORIES,
    "images": [],
    "annotations": []
}

image_id = 1
segmentation_id = 1

ROOT_DIR = "train"
IMAGE_DIR = "train/images"
ANNOTATION_DIR = "train/annotations"

image_files = [f for f in listdir(IMAGE_DIR) if isfile(join(IMAGE_DIR, f))]
annotation_files = [f for f in listdir(ANNOTATION_DIR) if isfile(join(ANNOTATION_DIR, f))]

# go through each image
for image_filename in image_files:
    image = Image.open(IMAGE_DIR + '/' + image_filename)
    image_info = pycococreatortools.create_image_info(image_id, os.path.basename(image_filename), image.size)

    coco_output["images"].append(image_info)

# go through each associated annotation
for annotation_filename in annotation_files:

    print(annotation_filename)
    class_id = [x['id'] for x in CATEGORIES if x['name'] in annotation_filename][0]
    category_info = {'id': class_id, 'is_crowd': 'crowd' in image_filename}
    binary_mask = np.asarray(Image.open(annotation_filename).convert('1')).astype(np.uint8)

    annotation_info = pycococreatortools.create_annotation_info(segmentation_id, image_id, category_info, binary_mask, image.size, tolerance=2)

    if annotation_info is not None:
        coco_output["annotations"].append(annotation_info)

    segmentation_id = segmentation_id + 1

    image_id = image_id + 1

with open('train/images.json', 'w') as output_json_file:
    json.dump(coco_output, output_json_file)

I still don't know if the code works or not. Please help!

numpy 1.18 change breaks pycocotools

this has been fixed in the original coco api repo, but just a heads up for anyone using pycocotools cocodataset/cocoapi#354

Extra annotations

Hi guys, does anyone have a problem adding extra annotations for some images?

[email protected] is not found

pip install git+git://github.com/waspinator/[email protected]
is not working

Question : Given Polygons Dots , Getting the area and the BoundingBox

I have the polygons dots that I labeled on an image and the image it self,

How can I estimate a Bounding Box and the area for the annotations ?

Thanks In advance

Error processing image,when use matterport Mask_RCNN

Hi,thanks for your codes. I use your code to produce .json file of my dataset. I have a problem when I use it for Mask RCNN. It may be fail to read the instance(mask) label. I don't know what's wrong about it. So I do need your suggestions,thanks.
I modify the coco.py to read my .json file and images.Then error below:

Epoch 1/40
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/008557.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [46.0, 194.0, 29.0, 20.0], 'category_id': 1, 'area': 496, 'image_id': 7355, 'segmentation': [[53.0, 213.5, 46.5, 211.0, 46.0, 195.5, 73.5, 195.0, 74.0, 210.5, 53.0, 213.5]], 'width': 640, 'height': 480, 'id': 74576}, {'iscrowd': 0, 'bbox': [299.0, 0.0, 235.0, 222.0], 'category_id': 23, 'area': 13517, 'image_id': 7355, 'segmentation': [[332.0, 205.5, 306.5, 203.0, 304.5, 49.0, 305.5, 19.0, 311.5, 18.0, 301.0, 19.5, 298.5, 15.0, 310.0, 14.5, 313.0, 11.5, 318.0, 13.5, 319.0, 10.5, 324.0, 12.5, 326.0, 9.5, 336.0, 11.5, 338.0, 8.5, 345.0, 10.5, 353.0, 9.5, 355.0, 6.5, 371.0, 7.5, 373.0, 4.5, 380.0, 6.5, 451.0, 0, 499.0, 32.5, 501.0, 35.5, 478.5, 46.0, 499.5, 75.0, 494.0, 95.5, 459.5, 54.0, 452.5, 35.0, 449.0, 5.5, 376.5, 14.0, 372.5, 42.0, 365.5, 60.0, 332.5, 104.0, 327.5, 117.0, 334.5, 165.0, 332.0, 205.5], [377.5, 11.0, 450.5, 3.0, 391.0, 7.5, 377.5, 11.0], [368.5, 12.0, 372.5, 11.0, 359.5, 12.0, 368.5, 12.0], [349.5, 14.0, 355.5, 13.0, 342.5, 14.0, 349.5, 14.0], [331.5, 16.0, 337.5, 14.0, 329.5, 15.0, 331.5, 16.0], [511.0, 221.5, 491.5, 218.0, 490.5, 183.0, 500.5, 132.0, 512.0, 116.5, 530.5, 128.0, 533.5, 170.0, 532.5, 186.0, 500.5, 203.0, 511.0, 221.5]], 'width': 640, 'height': 480, 'id': 74577}, {'iscrowd': 0, 'bbox': [198.0, 204.0, 353.0, 209.0], 'category_id': 10, 'area': 32022, 'image_id': 7355, 'segmentation': [[513.0, 412.5, 484.5, 405.0, 488.0, 330.5, 206.0, 279.5, 203.5, 278.0, 203.5, 262.0, 198.5, 260.0, 198.5, 250.0, 275.0, 229.5, 285.0, 230.5, 287.5, 209.0, 298.0, 203.5, 511.0, 221.5, 515.5, 227.0, 512.5, 260.0, 517.0, 257.5, 536.0, 259.5, 550.5, 282.0, 506.5, 368.0, 528.5, 383.0, 515.5, 403.0, 513.0, 412.5], [464.0, 381.5, 452.5, 379.0, 465.0, 365.5, 464.0, 381.5]], 'width': 640, 'height': 480, 'id': 74578}, {'iscrowd': 0, 'bbox': [0.0, 42.0, 188.0, 174.0], 'category_id': 12, 'area': 5764, 'image_id': 7355, 'segmentation': [[123.0, 215.5, 88.5, 209.0, 132.0, 202.5, 131.0, 183.5, 0, 197.0, 0, 186.0, 4.0, 184.5, 130.5, 173.0, 129.5, 144.0, 127.0, 140.5, 110.5, 140.0, 113.0, 137.5, 127.0, 137.5, 128.5, 134.0, 124.5, 68.0, 96.0, 64.5, 96.0, 73.5, 86.5, 73.0, 86.0, 83.5, 39.0, 80.5, 33.5, 77.0, 33.0, 68.5, 24.5, 67.0, 23.0, 57.5, 0.0, 55.5, 0.0, 41.5, 182.5, 63.0, 182.0, 66.5, 173.0, 67.5, 130.0, 67.5, 129.5, 72.0, 133.5, 134.0, 136.0, 137.5, 151.5, 138.0, 135.5, 141.0, 136.5, 172.0, 187.5, 175.0, 181.0, 178.5, 137.0, 182.5, 137.5, 201.0, 154.0, 199.5, 184.0, 202.5, 184.5, 205.0, 123.0, 215.5], [93.5, 71.0, 91.0, 63.5, 26.5, 58.0, 28.0, 66.5, 93.5, 71.0], [85.5, 81.0, 84.0, 72.5, 35.5, 69.0, 37.0, 78.5, 85.5, 81.0]], 'width': 640, 'height': 480, 'id': 74579}, {'iscrowd': 0, 'bbox': [2.0, 150.0, 43.0, 36.0], 'category_id': 1, 'area': 1318, 'image_id': 7355, 'segmentation': [[14.0, 184.5, 5.5, 183.0, 1.5, 151.0, 42.0, 150.5, 44.5, 181.0, 14.0, 184.5]], 'width': 640, 'height': 480, 'id': 74580}, {'iscrowd': 0, 'bbox': [63.0, 44.0, 90.0, 152.0], 'category_id': 20, 'area': 7111, 'image_id': 7355, 'segmentation': [[142.0, 58.5, 136.5, 57.0, 135.5, 48.0, 142.0, 47.5, 143.0, 43.5, 142.0, 58.5], [101.0, 53.5, 89.5, 52.0, 101.0, 50.5, 101.0, 53.5], [144.0, 58.5, 144.0, 51.5, 144.0, 58.5], [93.0, 71.5, 74.0, 70.5, 74.5, 63.0, 91.0, 63.5, 93.5, 65.0, 93.0, 71.5], [79.0, 178.5, 69.5, 178.0, 62.5, 108.0, 70.5, 83.0, 86.0, 83.5, 86.5, 73.0, 96.0, 73.5, 95.5, 65.0, 101.0, 64.5, 104.5, 66.0, 127.5, 112.0, 126.5, 137.0, 110.5, 139.0, 126.5, 141.0, 127.5, 173.0, 79.0, 178.5], [147.0, 137.5, 141.5, 137.0, 137.0, 68.5, 143.5, 70.0, 144.0, 78.5, 145.5, 70.0, 147.0, 137.5], [85.0, 81.5, 73.0, 79.5, 73.5, 72.0, 84.0, 72.5, 85.0, 81.5], [145.0, 172.5, 141.5, 141.0, 147.0, 139.5, 148.0, 148.5, 149.5, 141.0, 150.5, 171.0, 145.0, 172.5], [152.0, 195.5, 137.5, 194.0, 143.5, 192.0, 145.0, 181.5, 151.5, 182.0, 151.5, 189.0, 148.5, 192.0, 152.0, 195.5], [132.0, 194.5, 80.5, 189.0, 129.0, 183.5, 132.0, 194.5], [116.5, 189.0, 113.5, 189.0, 116.5, 189.0]], 'width': 640, 'height': 480, 'id': 74581}, {'iscrowd': 0, 'bbox': [0.0, 205.0, 194.0, 130.0], 'category_id': 26, 'area': 16028, 'image_id': 7355, 'segmentation': [[33.0, 334.5, 30.5, 334.0, 28.5, 321.0, 24.0, 313.5, 0, 308.0, 0.0, 234.5, 187.5, 205.0, 183.5, 229.0, 191.5, 234.0, 193.0, 276.5, 179.0, 276.5, 84.0, 305.5, 84.0, 317.5, 33.0, 334.5]], 'width': 640, 'height': 480, 'id': 74582}, {'iscrowd': 0, 'bbox': [0.0, 64.0, 23.0, 68.0], 'category_id': 8, 'area': 1307, 'image_id': 7355, 'segmentation': [[22.0, 131.5, 0, 131.0, 0.0, 63.5, 15.5, 65.0, 22.0, 131.5]], 'width': 640, 'height': 480, 'id': 74583}, {'iscrowd': 0, 'bbox': [0.0, 209.0, 119.0, 26.0], 'category_id': 13, 'area': 1344, 'image_id': 7355, 'segmentation': [[4.0, 234.5, 0, 234.0, 0.0, 220.5, 82.0, 208.5, 118.5, 214.0, 4.0, 234.5]], 'width': 640, 'height': 480, 'id': 74584}, {'iscrowd': 0, 'bbox': [0.0, 157.0, 7.0, 29.0], 'category_id': 1, 'area': 90, 'image_id': 7355, 'segmentation': [[1.0, 159.5, 0.0, 156.5, 1.0, 159.5], [1.0, 185.5, 0, 171.0, 4.0, 170.5, 6.5, 184.0, 1.0, 185.5]], 'width': 640, 'height': 480, 'id': 74585}, {'iscrowd': 0, 'bbox': [184.0, 182.0, 75.0, 66.0], 'category_id': 1, 'area': 3187, 'image_id': 7355, 'segmentation': [[209.0, 247.5, 205.5, 247.0, 204.0, 240.5, 183.5, 229.0, 187.5, 217.0, 185.5, 209.0, 194.5, 199.0, 193.5, 194.0, 200.0, 188.5, 202.0, 191.5, 201.5, 188.0, 205.0, 185.5, 214.0, 181.5, 216.0, 184.5, 232.0, 184.5, 232.5, 189.0, 241.0, 191.5, 247.5, 198.0, 247.5, 204.0, 250.5, 203.0, 252.5, 210.0, 249.5, 219.0, 257.0, 214.5, 258.5, 217.0, 235.0, 240.5, 209.0, 247.5]], 'width': 640, 'height': 480, 'id': 74586}, {'iscrowd': 0, 'bbox': [0.0, 191.0, 72.0, 30.0], 'category_id': 8, 'area': 1102, 'image_id': 7355, 'segmentation': [[5.0, 220.5, 0, 220.0, 0.0, 197.5, 63.0, 190.5, 71.5, 192.0, 46.0, 195.5, 47.0, 214.5, 5.0, 220.5]], 'width': 640, 'height': 480, 'id': 74587}, {'iscrowd': 0, 'bbox': [328.0, 13.0, 183.0, 206.0], 'category_id': 20, 'area': 26567, 'image_id': 7355, 'segmentation': [[449.0, 13.5, 446.5, 13.0, 449.0, 13.5], [439.0, 14.5, 436.5, 14.0, 439.0, 14.5], [429.0, 15.5, 426.5, 15.0, 429.0, 15.5], [419.0, 16.5, 416.5, 16.0, 419.0, 16.5], [409.0, 17.5, 406.5, 17.0, 409.0, 17.5], [399.0, 18.5, 396.5, 18.0, 399.0, 18.5], [491.0, 218.5, 396.0, 212.5, 334.5, 205.0, 334.5, 102.0, 365.5, 60.0, 376.0, 24.5, 450.0, 17.5, 453.5, 39.0, 459.5, 54.0, 494.5, 96.0, 490.5, 105.0, 510.5, 116.0, 500.5, 132.0, 493.5, 160.0, 490.5, 183.0, 491.0, 218.5], [389.0, 19.5, 386.5, 19.0, 389.0, 19.5], [379.0, 20.5, 376.5, 20.0, 379.0, 20.5], [328.0, 119.5, 328.0, 116.5, 328.0, 119.5]], 'width': 640, 'height': 480, 'id': 74588}, {'iscrowd': 0, 'bbox': [84.0, 276.0, 406.0, 204.0], 'category_id': 6, 'area': 41269, 'image_id': 7355, 'segmentation': [[440.0, 479.5, 436.5, 479.0, 438.5, 434.0, 436.0, 427.5, 399.0, 417.5, 386.0, 424.5, 359.0, 448.5, 220.0, 403.5, 217.5, 378.0, 211.0, 365.5, 183.0, 357.5, 158.0, 368.5, 154.5, 350.0, 137.5, 345.0, 161.5, 334.0, 159.0, 332.5, 117.0, 321.5, 118.5, 349.0, 88.0, 360.5, 83.5, 306.0, 188.0, 275.5, 488.0, 330.5, 482.5, 444.0, 472.0, 459.5, 460.5, 456.0, 462.5, 407.0, 460.0, 406.5, 443.5, 427.0, 440.0, 479.5], [464.5, 385.0, 465.0, 365.5, 449.5, 382.0, 464.5, 385.0]], 'width': 640, 'height': 480, 'id': 74589}, {'iscrowd': 0, 'bbox': [0.0, 309.0, 158.0, 133.0], 'category_id': 10, 'area': 6296, 'image_id': 7355, 'segmentation': [[0.0, 441.5, 0.0, 308.5, 14.0, 309.5, 26.5, 316.0, 33.0, 355.5, 42.0, 354.5, 69.0, 365.5, 77.0, 365.5, 130.0, 344.5, 148.0, 347.5, 156.5, 353.0, 156.0, 369.5, 117.0, 387.5, 89.0, 377.5, 71.0, 375.5, 8.0, 401.5, 4.5, 412.0, 10.5, 436.0, 0.0, 441.5]], 'width': 640, 'height': 480, 'id': 74590}, {'iscrowd': 0, 'bbox': [41.0, 33.0, 104.0, 160.0], 'category_id': 23, 'area': 4363, 'image_id': 7355, 'segmentation': [[136.0, 57.5, 102.0, 53.5, 100.0, 42.5, 74.0, 45.5, 74.0, 50.5, 42.0, 46.5, 40.5, 44.0, 137.0, 32.5, 144.5, 38.0, 135.5, 39.0, 136.0, 57.5], [135.5, 37.0, 131.5, 37.0, 135.5, 37.0], [107.5, 40.0, 104.5, 40.0, 107.5, 40.0], [80.5, 43.0, 99.5, 41.0, 75.5, 43.0, 80.5, 43.0], [73.0, 69.5, 46.0, 67.5, 45.5, 62.0, 50.0, 59.5, 74.0, 62.5, 73.0, 69.5], [126.0, 109.5, 106.5, 74.0, 105.0, 65.5, 124.5, 68.0, 126.0, 109.5], [141.0, 137.5, 136.0, 137.5, 133.5, 134.0, 130.0, 67.5, 136.0, 68.5, 137.5, 72.0, 141.0, 137.5], [72.0, 79.5, 48.0, 78.5, 45.5, 70.0, 73.0, 71.5, 72.0, 79.5], [57.0, 180.5, 49.0, 80.5, 70.5, 83.0, 62.5, 108.0, 69.5, 178.0, 57.0, 180.5], [127.0, 137.5, 128.0, 127.5, 127.0, 137.5], [130.0, 173.5, 127.5, 173.0, 126.5, 164.0, 127.0, 140.5, 129.5, 144.0, 130.0, 173.5], [143.0, 172.5, 135.5, 170.0, 136.0, 140.5, 141.5, 141.0, 143.0, 172.5], [143.0, 192.5, 137.5, 192.0, 137.0, 182.5, 143.5, 183.0, 143.0, 192.5], [132.0, 191.5, 129.5, 191.0, 130.0, 183.5, 132.0, 191.5]], 'width': 640, 'height': 480, 'id': 74591}, {'iscrowd': 0, 'bbox': [452.0, 0.0, 188.0, 480.0], 'category_id': 1, 'area': 55519, 'image_id': 7355, 'segmentation': [[639.0, 479.5, 561.0, 479.5, 543.5, 461.0, 578.5, 418.0, 506.5, 368.0, 550.5, 282.0, 500.5, 203.0, 539.0, 184.5, 540.5, 179.0, 529.5, 126.0, 490.5, 105.0, 493.5, 102.0, 499.5, 75.0, 478.5, 46.0, 501.5, 35.0, 452.0, 0, 639.5, 0.0, 639.0, 479.5]], 'width': 640, 'height': 480, 'id': 74592}, {'iscrowd': 0, 'bbox': [0.0, 358.0, 439.0, 122.0], 'category_id': 10, 'area': 32061, 'image_id': 7355, 'segmentation': [[436.0, 479.5, 0.0, 479.5, 0, 442.0, 10.5, 436.0, 4.5, 412.0, 8.0, 401.5, 71.0, 375.5, 89.0, 377.5, 117.0, 387.5, 178.0, 358.5, 188.0, 357.5, 208.0, 363.5, 213.5, 368.0, 218.5, 384.0, 221.0, 418.5, 235.0, 417.5, 265.0, 427.5, 274.0, 433.5, 286.0, 434.5, 319.0, 445.5, 331.0, 453.5, 335.0, 451.5, 353.0, 457.5, 386.0, 424.5, 399.0, 417.5, 436.0, 427.5, 438.5, 434.0, 436.0, 479.5]], 'width': 640, 'height': 480, 'id': 74593}, {'iscrowd': 0, 'bbox': [0.0, 0.0, 96.0, 16.0], 'category_id': 34, 'area': 807, 'image_id': 7355, 'segmentation': [[4.0, 15.5, 0, 15.0, 0.0, 0, 95.5, 0.0, 4.0, 15.5]], 'width': 640, 'height': 480, 'id': 74594}], 'width': 640, 'height': 480, 'id': 7355}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/016359.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [0.0, 431.0, 516.0, 49.0], 'category_id': 30, 'area': 2033, 'image_id': 8885, 'segmentation': [[0.0, 471.5, 0.0, 430.5, 43.5, 435.0, 0.0, 471.5], [515.0, 479.5, 368.0, 479.5, 367.5, 466.0, 515.0, 479.5]], 'width': 640, 'height': 480, 'id': 90126}, {'iscrowd': 0, 'bbox': [580.0, 140.0, 60.0, 287.0], 'category_id': 12, 'area': 8117, 'image_id': 8885, 'segmentation': [[597.0, 426.5, 590.5, 425.0, 589.5, 417.0, 592.5, 396.0, 590.5, 377.0, 592.5, 359.0, 596.5, 355.0, 597.5, 336.0, 579.5, 319.0, 594.5, 144.0, 596.0, 139.5, 639.0, 139.5, 639.0, 150.5, 626.0, 150.5, 624.5, 153.0, 620.5, 201.0, 639.0, 201.5, 639.5, 220.0, 618.5, 220.0, 614.5, 257.0, 639.0, 257.5, 639.5, 282.0, 612.0, 281.5, 608.5, 311.0, 639.5, 314.0, 639.5, 342.0, 606.0, 340.5, 603.5, 360.0, 639.5, 362.0, 639.5, 398.0, 601.0, 394.5, 597.0, 426.5], [611.5, 201.0, 616.5, 148.0, 603.0, 147.5, 597.5, 200.0, 611.5, 201.0], [599.5, 257.0, 605.0, 256.5, 606.5, 253.0, 609.5, 216.0, 597.0, 209.5, 592.5, 256.0, 599.5, 257.0], [594.5, 311.0, 600.5, 310.0, 603.5, 278.0, 592.0, 267.5, 587.5, 309.0, 594.5, 311.0], [627.5, 380.0, 625.5, 379.0, 627.5, 380.0], [614.5, 383.0, 612.5, 382.0, 614.5, 383.0], [631.5, 385.0, 629.5, 384.0, 631.5, 385.0], [617.5, 387.0, 615.5, 386.0, 617.5, 387.0]], 'width': 640, 'height': 480, 'id': 90127}, {'iscrowd': 0, 'bbox': [0.0, 395.0, 370.0, 85.0], 'category_id': 6, 'area': 24783, 'image_id': 8885, 'segmentation': [[367.0, 479.5, 0.0, 479.5, 0, 472.0, 88.0, 397.5, 92.0, 394.5, 102.0, 394.5, 369.0, 416.5, 367.0, 479.5]], 'width': 640, 'height': 480, 'id': 90128}, {'iscrowd': 0, 'bbox': [74.0, 0.0, 32.0, 32.0], 'category_id': 7, 'area': 697, 'image_id': 8885, 'segmentation': [[98.0, 31.5, 73.5, 27.0, 75.5, 6.0, 82.0, 0, 86.0, 4.5, 85.5, 0.0, 88.0, 0, 90.5, 10.0, 95.0, 8.5, 104.5, 16.0, 104.5, 26.0, 98.0, 31.5]], 'width': 640, 'height': 480, 'id': 90129}, {'iscrowd': 0, 'bbox': [426.0, 306.0, 169.0, 119.0], 'category_id': 5, 'area': 16338, 'image_id': 8885, 'segmentation': [[586.0, 424.5, 582.5, 424.0, 578.0, 414.5, 436.0, 403.5, 430.0, 403.5, 429.5, 412.0, 427.0, 412.5, 425.5, 364.0, 428.5, 306.0, 575.0, 313.5, 594.5, 337.0, 586.0, 424.5]], 'width': 640, 'height': 480, 'id': 90130}, {'iscrowd': 0, 'bbox': [503.0, 0.0, 26.0, 20.0], 'category_id': 7, 'area': 457, 'image_id': 8885, 'segmentation': [[517.0, 19.5, 509.0, 19.5, 502.5, 13.0, 502.5, 5.0, 506.0, 0, 527.5, 0.0, 528.5, 15.0, 517.0, 19.5]], 'width': 640, 'height': 480, 'id': 90131}, {'iscrowd': 0, 'bbox': [0.0, 137.0, 61.0, 243.0], 'category_id': 12, 'area': 6372, 'image_id': 8885, 'segmentation': [[31.0, 379.5, 25.5, 379.0, 22.0, 352.5, 0, 350.0, 0, 328.0, 9.5, 323.0, 7.5, 306.0, 0, 305.0, 0, 284.0, 4.5, 282.0, 2.5, 259.0, 0, 255.0, 0.0, 170.5, 1.5, 188.0, 5.5, 188.0, 0.0, 136.5, 47.5, 137.0, 60.5, 289.0, 25.5, 304.0, 31.5, 340.0, 33.5, 377.0, 31.0, 379.5], [44.5, 189.0, 40.5, 144.0, 10.0, 143.5, 13.5, 188.0, 44.5, 189.0], [48.5, 236.0, 45.5, 197.0, 17.0, 200.5, 18.5, 235.0, 48.5, 236.0], [10.5, 235.0, 7.5, 204.0, 4.0, 203.5, 6.5, 235.0, 10.5, 235.0], [53.5, 282.0, 49.0, 244.5, 20.5, 254.0, 24.0, 281.5, 53.5, 282.0], [12.5, 281.0, 15.5, 278.0, 13.5, 259.0, 10.0, 255.5, 12.5, 281.0], [19.5, 322.0, 18.5, 307.0, 15.0, 306.5, 15.5, 320.0, 19.5, 322.0]], 'width': 640, 'height': 480, 'id': 90132}, {'iscrowd': 0, 'bbox': [169.0, 195.0, 259.0, 207.0], 'category_id': 2, 'area': 45540, 'image_id': 8885, 'segmentation': [[425.0, 401.5, 174.0, 382.5, 172.5, 376.0, 186.5, 362.0, 186.5, 353.0, 181.5, 247.0, 180.5, 238.0, 175.5, 237.0, 180.5, 233.0, 179.5, 210.0, 168.5, 202.0, 190.0, 194.5, 426.0, 199.5, 427.5, 209.0, 415.5, 216.0, 414.5, 232.0, 420.5, 233.0, 420.5, 245.0, 414.5, 246.0, 410.5, 354.0, 410.5, 370.0, 425.5, 372.0, 425.0, 401.5]], 'width': 640, 'height': 480, 'id': 90133}, {'iscrowd': 0, 'bbox': [28.0, 287.0, 159.0, 102.0], 'category_id': 5, 'area': 13212, 'image_id': 8885, 'segmentation': [[162.0, 388.5, 158.5, 388.0, 158.5, 381.0, 153.0, 379.5, 60.0, 372.5, 47.0, 372.5, 36.0, 379.5, 27.5, 305.0, 65.0, 286.5, 183.0, 292.5, 186.5, 362.0, 162.0, 388.5]], 'width': 640, 'height': 480, 'id': 90134}, {'iscrowd': 0, 'bbox': [187.0, 0.0, 230.0, 83.0], 'category_id': 25, 'area': 18564, 'image_id': 8885, 'segmentation': [[242.0, 82.5, 189.5, 81.0, 187.0, 0, 416.5, 0.0, 414.5, 71.0, 411.0, 80.5, 242.0, 82.5]], 'width': 640, 'height': 480, 'id': 90135}], 'width': 640, 'height': 480, 'id': 8885}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/000418.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [529.0, 462.0, 109.0, 18.0], 'category_id': 1, 'area': 1535, 'image_id': 5581, 'segmentation': [[637.0, 479.5, 531.0, 479.5, 528.5, 464.0, 582.0, 462.5, 623.0, 470.5, 635.0, 474.5, 637.0, 479.5]], 'width': 640, 'height': 480, 'id': 56444}, {'iscrowd': 0, 'bbox': [520.0, 190.0, 82.0, 247.0], 'category_id': 25, 'area': 13836, 'image_id': 5581, 'segmentation': [[577.0, 436.5, 568.5, 433.0, 555.5, 401.0, 537.5, 397.0, 525.5, 365.0, 526.0, 357.5, 536.5, 357.0, 519.5, 319.0, 530.0, 189.5, 536.0, 190.5, 592.0, 228.5, 601.5, 242.0, 586.5, 407.0, 585.5, 412.0, 580.5, 413.0, 577.0, 436.5]], 'width': 640, 'height': 480, 'id': 56445}, {'iscrowd': 0, 'bbox': [479.0, 323.0, 150.0, 157.0], 'category_id': 5, 'area': 11620, 'image_id': 5581, 'segmentation': [[530.0, 479.5, 491.0, 479.5, 489.5, 476.0, 478.5, 406.0, 483.0, 322.5, 522.5, 325.0, 536.5, 357.0, 526.0, 357.5, 525.5, 365.0, 537.5, 397.0, 555.5, 401.0, 572.0, 436.5, 578.5, 435.0, 580.5, 413.0, 585.5, 412.0, 587.0, 397.5, 628.5, 472.0, 582.0, 462.5, 547.0, 461.5, 528.5, 464.0, 530.0, 479.5]], 'width': 640, 'height': 480, 'id': 56446}, {'iscrowd': 0, 'bbox': [117.0, 352.0, 296.0, 128.0], 'category_id': 30, 'area': 28619, 'image_id': 5581, 'segmentation': [[402.0, 479.5, 116.5, 479.0, 208.5, 377.0, 208.0, 351.5, 412.0, 367.5, 402.0, 479.5]], 'width': 640, 'height': 480, 'id': 56447}, {'iscrowd': 0, 'bbox': [0.0, 309.0, 209.0, 171.0], 'category_id': 14, 'area': 29287, 'image_id': 5581, 'segmentation': [[116.0, 479.5, 0, 479.0, 0.0, 308.5, 158.0, 319.5, 179.0, 322.5, 187.0, 327.5, 199.0, 317.5, 205.5, 317.0, 208.5, 377.0, 116.0, 479.5]], 'width': 640, 'height': 480, 'id': 56448}, {'iscrowd': 0, 'bbox': [62.0, 37.0, 127.0, 254.0], 'category_id': 15, 'area': 15151, 'image_id': 5581, 'segmentation': [[188.0, 290.5, 81.5, 284.0, 62.0, 39.5, 177.5, 37.0, 188.0, 290.5], [174.5, 281.0, 147.5, 264.0, 142.5, 176.0, 141.5, 171.0, 136.5, 170.0, 141.5, 168.0, 141.5, 159.0, 135.5, 57.0, 162.5, 49.0, 79.0, 50.5, 95.5, 276.0, 174.5, 281.0]], 'width': 640, 'height': 480, 'id': 56449}, {'iscrowd': 0, 'bbox': [564.0, 0.0, 76.0, 348.0], 'category_id': 20, 'area': 16241, 'image_id': 5581, 'segmentation': [[565.0, 210.5, 563.5, 198.0, 579.5, 4.0, 582.0, 0, 565.0, 210.5], [639.0, 309.5, 599.5, 273.0, 601.5, 242.0, 592.0, 228.5, 575.5, 218.0, 595.0, 0, 639.5, 0.0, 639.0, 309.5], [639.0, 347.5, 596.5, 301.0, 598.0, 284.5, 639.5, 326.0, 639.0, 347.5]], 'width': 640, 'height': 480, 'id': 56450}], 'width': 640, 'height': 480, 'id': 5581}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/014976.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [0.0, 82.0, 145.0, 386.0], 'category_id': 5, 'area': 39708, 'image_id': 3308, 'segmentation': [[0.0, 467.5, 0, 91.0, 34.0, 92.5, 43.0, 83.5, 57.0, 81.5, 68.0, 85.5, 76.0, 93.5, 130.0, 96.5, 138.0, 94.5, 144.5, 105.0, 141.0, 115.5, 103.5, 115.0, 116.5, 284.0, 116.0, 289.5, 104.5, 292.0, 113.5, 419.0, 0.0, 467.5]], 'width': 640, 'height': 480, 'id': 33090}, {'iscrowd': 0, 'bbox': [446.0, 15.0, 194.0, 465.0], 'category_id': 15, 'area': 87855, 'image_id': 3308, 'segmentation': [[639.0, 479.5, 445.5, 479.0, 468.0, 0, 639.5, 0.0, 639.0, 479.5]], 'width': 640, 'height': 480, 'id': 33091}, {'iscrowd': 0, 'bbox': [167.0, 277.0, 125.0, 203.0], 'category_id': 24, 'area': 16455, 'image_id': 3308, 'segmentation': [[262.0, 479.5, 189.0, 479.5, 175.5, 471.0, 167.0, 284.5, 199.0, 276.5, 222.0, 280.5, 225.5, 374.0, 239.0, 372.5, 269.0, 384.5, 290.5, 401.0, 291.5, 432.0, 278.5, 456.0, 262.5, 474.0, 262.0, 479.5]], 'width': 640, 'height': 480, 'id': 33092}, {'iscrowd': 0, 'bbox': [0.0, 0.0, 177.0, 480.0], 'category_id': 23, 'area': 39050, 'image_id': 3308, 'segmentation': [[176.0, 479.5, 0, 479.0, 1.0, 466.5, 113.5, 419.0, 104.5, 292.0, 116.0, 289.5, 116.5, 284.0, 103.5, 115.0, 139.0, 116.5, 144.5, 107.0, 143.5, 100.0, 138.0, 94.5, 130.0, 96.5, 76.0, 93.5, 68.0, 85.5, 57.0, 81.5, 43.0, 83.5, 34.0, 92.5, 0.0, 90.5, 0.0, 0, 150.5, 0.0, 176.0, 479.5]], 'width': 640, 'height': 480, 'id': 33093}, {'iscrowd': 0, 'bbox': [253.0, 361.0, 3.0, 19.0], 'category_id': 1, 'area': 54, 'image_id': 3308, 'segmentation': [[255.0, 379.5, 252.5, 378.0, 253.0, 360.5, 255.5, 362.0, 255.0, 379.5]], 'width': 640, 'height': 480, 'id': 33094}], 'width': 640, 'height': 480, 'id': 3308}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/015192.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [0.0, 33.0, 142.0, 189.0], 'category_id': 20, 'area': 22160, 'image_id': 7137, 'segmentation': [[6.0, 221.5, 0, 221.0, 0, 37.0, 123.0, 48.5, 133.5, 200.0, 128.5, 203.0, 139.0, 202.5, 141.5, 205.0, 6.0, 221.5], [130.0, 64.5, 130.0, 54.5, 130.0, 64.5], [131.0, 81.5, 131.0, 71.5, 131.0, 81.5], [132.0, 97.5, 132.0, 87.5, 132.0, 97.5], [133.0, 114.5, 133.0, 104.5, 133.0, 114.5], [134.0, 130.5, 134.0, 120.5, 134.0, 130.5], [135.0, 146.5, 135.0, 137.5, 135.0, 146.5], [136.0, 163.5, 136.0, 153.5, 136.0, 163.5], [137.0, 179.5, 137.0, 169.5, 137.0, 179.5], [138.0, 196.5, 138.0, 186.5, 138.0, 196.5], [127.5, 204.0, 119.5, 204.0, 127.5, 204.0], [118.5, 205.0, 110.5, 205.0, 118.5, 205.0], [92.5, 208.0, 109.5, 206.0, 88.5, 208.0, 92.5, 208.0], [83.5, 209.0, 87.5, 208.0, 80.5, 209.0, 83.5, 209.0], [74.5, 210.0, 76.5, 209.0, 74.5, 210.0], [65.5, 211.0, 67.5, 210.0, 65.5, 211.0]], 'width': 640, 'height': 480, 'id': 72306}, {'iscrowd': 0, 'bbox': [183.0, 0.0, 202.0, 19.0], 'category_id': 34, 'area': 2075, 'image_id': 7137, 'segmentation': [[277.0, 18.5, 182.5, 3.0, 187.0, 0, 227.0, 0, 244.0, 3.5, 252.0, 0, 384.5, 0.0, 277.0, 18.5]], 'width': 640, 'height': 480, 'id': 72307}, {'iscrowd': 0, 'bbox': [32.0, 300.0, 608.0, 180.0], 'category_id': 1, 'area': 68029, 'image_id': 7137, 'segmentation': [[639.0, 479.5, 31.5, 479.0, 358.0, 338.5, 364.5, 335.0, 366.0, 329.5, 373.0, 326.5, 381.0, 328.5, 449.0, 299.5, 639.0, 342.5, 639.0, 479.5]], 'width': 640, 'height': 480, 'id': 72308}, {'iscrowd': 0, 'bbox': [564.0, 63.0, 76.0, 280.0], 'category_id': 14, 'area': 15425, 'image_id': 7137, 'segmentation': [[639.0, 342.5, 563.5, 325.0, 577.5, 157.0, 599.5, 156.0, 606.5, 77.0, 639.0, 62.5, 639.0, 342.5]], 'width': 640, 'height': 480, 'id': 72309}, {'iscrowd': 0, 'bbox': [327.0, 305.0, 96.0, 39.0], 'category_id': 30, 'area': 1331, 'image_id': 7137, 'segmentation': [[346.0, 343.5, 330.0, 338.5, 326.5, 333.0, 396.0, 304.5, 422.5, 310.0, 381.0, 328.5, 373.0, 326.5, 366.0, 329.5, 364.0, 335.5, 346.0, 343.5]], 'width': 640, 'height': 480, 'id': 72310}, {'iscrowd': 0, 'bbox': [158.0, 0.0, 94.0, 7.0], 'category_id': 1, 'area': 181, 'image_id': 7137, 'segmentation': [[161.0, 6.5, 158.0, 0, 186.5, 0.0, 185.0, 2.5, 161.0, 6.5], [246.0, 3.5, 227.5, 0.0, 251.5, 0.0, 246.0, 3.5]], 'width': 640, 'height': 480, 'id': 72311}, {'iscrowd': 0, 'bbox': [363.0, 46.0, 75.0, 94.0], 'category_id': 26, 'area': 6487, 'image_id': 7137, 'segmentation': [[433.0, 139.5, 362.5, 137.0, 364.0, 53.5, 437.5, 47.0, 433.0, 139.5]], 'width': 640, 'height': 480, 'id': 72312}], 'width': 640, 'height': 480, 'id': 7137}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
ERROR:root:Error processing image {'source': 'coco', 'path': '/home/he/sundata/SUNCGxS_IMG/train/color/008931.jpg', 'annotations': [{'iscrowd': 0, 'bbox': [301.0, 290.0, 242.0, 190.0], 'category_id': 24, 'area': 13430, 'image_id': 3614, 'segmentation': [[539.0, 363.5, 467.5, 356.0, 472.0, 292.5, 487.0, 289.5, 503.0, 290.5, 522.0, 292.5, 531.5, 297.0, 541.5, 328.0, 541.5, 347.0, 540.0, 351.5, 523.0, 350.5, 522.5, 354.0, 539.0, 356.5, 539.0, 363.5], [375.0, 479.5, 336.5, 479.0, 343.5, 451.0, 326.5, 434.0, 300.5, 391.0, 303.5, 371.0, 309.0, 364.5, 328.0, 357.5, 348.0, 355.5, 375.0, 355.5, 408.5, 360.0, 375.0, 479.5]], 'width': 640, 'height': 480, 'id': 36061}, {'iscrowd': 0, 'bbox': [376.0, 349.0, 221.0, 131.0], 'category_id': 18, 'area': 22881, 'image_id': 3614, 'segmentation': [[596.0, 479.5, 376.0, 479.5, 410.5, 354.0, 418.0, 351.5, 539.0, 363.5, 539.0, 356.5, 525.0, 355.5, 522.5, 351.0, 556.0, 348.5, 567.0, 352.5, 565.5, 389.0, 596.0, 479.5]], 'width': 640, 'height': 480, 'id': 36062}, {'iscrowd': 0, 'bbox': [174.0, 417.0, 145.0, 63.0], 'category_id': 30, 'area': 6618, 'image_id': 3614, 'segmentation': [[297.0, 479.5, 173.5, 479.0, 226.0, 418.5, 249.0, 416.5, 307.0, 424.5, 318.5, 431.0, 318.5, 437.0, 297.0, 479.5]], 'width': 640, 'height': 480, 'id': 36063}, {'iscrowd': 0, 'bbox': [298.0, 251.0, 220.0, 117.0], 'category_id': 22, 'area': 15439, 'image_id': 3614, 'segmentation': [[306.0, 367.5, 297.5, 352.0, 297.5, 283.0, 301.5, 273.0, 325.0, 251.5, 377.0, 251.5, 369.0, 253.5, 365.5, 262.0, 377.0, 272.5, 386.5, 276.0, 383.5, 284.0, 388.5, 289.0, 373.5, 303.0, 383.0, 306.5, 402.0, 304.5, 411.0, 292.5, 436.0, 290.5, 452.0, 281.5, 459.5, 275.0, 468.0, 257.5, 510.5, 261.0, 517.0, 292.5, 487.0, 289.5, 474.0, 291.5, 470.5, 295.0, 467.0, 356.5, 418.0, 351.5, 413.0, 351.5, 409.0, 359.5, 375.0, 355.5, 333.0, 356.5, 317.0, 360.5, 306.0, 367.5], [413.0, 273.5, 409.5, 272.0, 415.0, 270.5, 413.0, 273.5]], 'width': 640, 'height': 480, 'id': 36064}, {'iscrowd': 0, 'bbox': [366.0, 150.0, 151.0, 157.0], 'category_id': 21, 'area': 9389, 'image_id': 3614, 'segmentation': [[388.0, 306.5, 373.5, 304.0, 388.5, 289.0, 383.5, 284.0, 386.5, 276.0, 373.0, 269.5, 365.5, 262.0, 365.5, 257.0, 371.0, 252.5, 392.0, 249.5, 416.0, 230.5, 423.0, 228.5, 437.5, 205.0, 455.0, 187.5, 465.0, 184.5, 485.0, 186.5, 481.5, 183.0, 490.5, 154.0, 502.0, 149.5, 511.0, 152.5, 516.5, 161.0, 515.5, 169.0, 495.5, 191.0, 512.5, 212.0, 511.5, 220.0, 500.5, 242.0, 487.0, 253.5, 469.0, 256.5, 459.5, 275.0, 452.0, 281.5, 436.0, 290.5, 411.0, 292.5, 402.0, 304.5, 388.0, 306.5], [428.5, 250.0, 442.5, 239.0, 450.0, 218.5, 421.0, 242.5, 402.5, 248.0, 428.5, 250.0], [481.5, 242.0, 482.0, 239.5, 481.5, 242.0], [413.5, 273.0, 415.0, 270.5, 409.5, 272.0, 413.5, 273.0]], 'width': 640, 'height': 480, 'id': 36065}], 'width': 640, 'height': 480, 'id': 3614}
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'
Exception in thread Thread-2:
Traceback (most recent call last):
  File "/usr/lib/python3.5/threading.py", line 914, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.5/threading.py", line 862, in run
    self._target(*self._args, **self._kwargs)
  File "/usr/local/lib/python3.5/dist-packages/keras/utils/data_utils.py", line 568, in data_generator_task
    generator_output = next(self._generator)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1710, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 1276, in load_image_gt
    source_class_ids = dataset.source_class_ids[dataset.image_info[image_id]["source"]]
KeyError: 'coco'

Traceback (most recent call last):
  File "samples/house/house.py", line 428, in <module>
    augmentation=augmentation)
  File "/usr/local/lib/python3.5/dist-packages/mask_rcnn-2.1-py3.5.egg/mrcnn/model.py", line 2375, in train
  File "/usr/local/lib/python3.5/dist-packages/keras/legacy/interfaces.py", line 87, in wrapper
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/keras/engine/training.py", line 2011, in fit_generator
    generator_output = next(output_generator)
StopIteration

```

There should be an option to create RLE or uncompressed RLE in create_annotation_info independent of "is_crowd".

@waspinator For example, for coco stuff task, they use RLE format segmentation and is_crowd is set to 0 or unset. If the segmentation in a custom dataset need to be represented by RLE, then it should be independent of "is_crowd", since the evaluation of iou would be wrong in when "is_crowd=True".

Panoptic Polygons Overlapping

I've been trying to build polygons for a model that requires me to convert them to panoptic format. I've been running into issues because I can't make everything a crowd, but if I don't, I get polygons that overlap. I should also note that I have double and triple checked (going as far as to subtract all masks from each other). Is there any way around this?

skim age to scikit-image

Hi, thanks for this tool! Can you please change skimage to scikit-image so I can just pip install git+this_repo?

NameError: name 'pycococreatortools' is not defined

I have installed using this:

!pip install cython
!pip install git+git://github.com/waspinator/[email protected]

But still I got this error. Whole code:

INFO = {
    "description": "Fashion Dataset",
    "url": "https://github.com/waspinator/pycococreator",
    "version": "0.1.0",
    "year": 2020,
    "contributor": "Abu Noman Md. Sakib",
    "date_created": datetime.datetime.utcnow().isoformat(' ')
}

LICENSES = [
    {
        "id": 1,
        "name": "GB",
        "url": "GB"
    }
]

CATEGORIES = [
    {
        'id': 1,
        'name': 'mask',
        'supercategory': 'fashion',
    }
]

coco_output = {
    "info": INFO,
    "licenses": LICENSES,
    "categories": CATEGORIES,
    "images": [],
    "annotations": []
}

image_id = 1
segmentation_id = 1

ROOT_DIR = "train"
IMAGE_DIR = "train/images"
ANNOTATION_DIR = "train/annotations"

image_files = [f for f in listdir(IMAGE_DIR) if isfile(join(IMAGE_DIR, f))]
annotation_files = [f for f in listdir(ANNOTATION_DIR) if isfile(join(ANNOTATION_DIR, f))]

# go through each image
for image_filename in image_files:
    image = Image.open(IMAGE_DIR + '/' + image_filename)
    image_info = pycococreatortools.create_image_info(image_id, os.path.basename(image_filename), image.size)

    coco_output["images"].append(image_info)

# go through each associated annotation
for annotation_filename in annotation_files:

    print(annotation_filename)
    class_id = [x['id'] for x in CATEGORIES if x['name'] in annotation_filename][0]
    category_info = {'id': class_id, 'is_crowd': 'crowd' in image_filename}
    binary_mask = np.asarray(Image.open(annotation_filename).convert('1')).astype(np.uint8)

    annotation_info = pycococreatortools.create_annotation_info(segmentation_id, image_id, category_info, binary_mask, image.size, tolerance=2)

    if annotation_info is not None:
        coco_output["annotations"].append(annotation_info)

    segmentation_id = segmentation_id + 1

    image_id = image_id + 1

with open('train/images.json', 'w') as output_json_file:
    json.dump(coco_output, output_json_file)

I still don't know if the code works or not. Please help!

Bug: Mismatch of class names from images and predefined class names, if names have same word.

This line of code finds the wrong class id, if the names are "inherent".
https://github.com/waspinator/pycococreator/blob/master/examples/shapes/shapes_to_coco.py#L102

E.g. if the class name of the image has race car will that trigger the first option because car is a valid match.

CATEGORIES = [
    {
        'id': 1,
        'name': 'car',
        'supercategory': 'vehicles',
    },
    {
        'id': 2,
        'name': 'race car',
        'supercategory': 'vehicles',
    }
]

VGG annotation to COCO?

I have labeled my data using VGG annotator, which provide a Json file at the end. is there a possibility to convert that file into COCO format? thank you

Holes are filled

It seems to me that if the annotation sources contain holes, the generated json will fill them up.

How to solve multiple polygons(holes) in one instance segmentation?

As describe in question, I met one issue. For example, for a person's instance segmentation, there might be more than 1 polygon in the mask. some holes appeared in the person's polygon? How to solve this occasion? @waspinator

My Json is different from COCO

Hi everyone! I have successed achieve the file of json in my dataset, but the style is different as fellow:

It seems have more "" , can i resolve it? Thank you!

Why substracting 1 is needed for contours?

I can not understand why substracting in this line is needed. @waspinator Could you explain it?

Circle Annotation Bug

Hello,

when i use your tool, there is a bug.

If i have a rope which is arranged like a circle, your tool give me this output:

This is the full mask:

This is the mask which i use with your tool:

And this is when i inspect my data with the annotation file:

Do anybody know how i can solve this problem?

Pycococreator with mutiprocessing

Can we have a code where we can use power of multiprocessing with very large datasets because with present code its taking too much time to process it.

Thanks in advance!

Tutorial for part before pycococreater

HI,
Thank you for the post, it was really informative. I am new to this so i am having difficulty converting my dataset to COCO format. Specifically the part about using Annotator tool and Binary masks. Can you please explain this in detail.
Thanks

/annotations/instances_valminusminival2018.json file missing

Trying to train Mask R CNN, I get the error that above file is missing.
Did you try to use your created dataset for a model like Mask R CNN? Maybe there are more files missing. Eager to hear from you and thx.

install protocol needs to switch to https

git.github.com is no longer supported

https://stackoverflow.com/questions/70663523/the-unauthenticated-git-protocol-on-port-9418-is-no-longer-supported

pip install git+https://github.com/waspinator/[email protected] works

how to annotate an image like this?

dear all,

I would like to use pycococreator to annotate fish(es) from an image like https://www.rodfishingclub.com/wp-content/uploads/2016/08/Skipjack-tuna-or-Katsuwonus-pelamis-Rod-Fishing-Club-Rodrigues-Island-Mauritius-Indian-Ocean.jpg
how to start it? thank you very much in advance.

warmest regards,
suryadi

COCO sample previews show multiple sample masks

I'm not sure what the deal is, but I set up my dataset as described in your guide. The json file generates fine, but when I generate several random previews, some of my samples have some technicolor nonsense going on.

By that I mean 11 masks (they happen to be consecutive masks in the index) are drawn/assigned over the one sample. I've checked the json file, and it looks like all these annotations have been assigned to the one sample for some reason. Each sample is named according to the suggested format, so I'm not sure what's up.

For example:

image filename: DSC_5409_1.jpg
annotation filename: DSC_5409_1_mask_1.png

image: MVI_0155_1107_140.jpg
annotation: MVI_0155_1107_140_mask_1.png

There's the same number of unique filenames under each directory, and I've stripped the extensions and "_mask_1" from the annotation filenames to ensure they match.

Anyone else have this problem?

===

To explain further, I have a source image, let's say MVI_0155_1107.jpg. This is sampled using a sliding window technique to produce x number of samples saved as MVI_0155_1107_#.jpg, with x going as high as 350 in some cases.

...ah, that's the problem. When it reaches images xx_3.jpg, and starts looking for annotations, it's going to pick up ANY annotations with 'xx_3' in the filename. So, if I have annotations with: xx_3, xx_30, xx_31, xx_32, xx_33, xx_34, xx_35, xx_36, xx_37, xx_38, and xx_39, that leaves me with 11 annotations assigned to one sample.

So it's a matter of improving how the script parses annotation filenames for comparison. I'll try something like this...

#  I'm going to index out what I need.  To do that...

words_to_strip = annotation_filename.split('_')[-2:]  # in our case we'll get ['mask', '1.png']

char_sum = 0
for word in words_to_strip:
    char_sum += len(word)

# We still need to add 2 to account for the underscore before and after 'mask'.

char_sum +=2

# Use the sum to index out the part of the filename we want to compare

annotation_filename_match = annotation_filename[:-char_sum]

Now we can use the image filename to match to the appropriate annotation regardless of naming scheme.

There are probably better ways to do this, so please feel free to suggest! I'm not closing this yet because I haven't tested my method. :)

Error processing image, ValueError: Input image expected to be RGB, RGBA or gray.

I'm getting this error very intermittently during training (approx. 10 times per 1000 iterations). I have variable size images and masks, so I'm thinking this may be an issue with some of the very large images in my dataset (for example, sizes 5456x3632, 2592x1944, etc.). It continues to train without crashing due to the error, but I'm unsure if there will be any negative consequences later on.

ERROR:root:Error processing image {'id': 3452, 'source': 'coco', 'path': '/home/docker_user/data/typeb_data/train2018/00001207.jpg', 'width': 5456, 'height': 3632, 'annotations': [{'id': 8215, 'image_id': 3452, 'category_id': 1, 'iscrowd': 0, 'area': 2581761, 'bbox': [2971.0, 657.0, 2217.0, 2258.0], 'segmentation': [[4691.0, 2914.5, 4657.0, 2912.5, 4640.0, 2909.5, 4595.0, 2909.5, 4571.0, 2903.5, 4522.0, 2879.5, 4503.0, 2841.5, 4491.0, 2839.5, 4475.0, 2833.5, 4446.5, 2818.0, 4443.5, 2767.0, 4438.5, 2735.0, 4432.5, 2722.0, 4416.5, 2710.0, 4411.5, 2696.0, 4400.5, 2676.0, 4397.5, 2662.0, 4400.5, 2642.0, 4408.0, 2632.5, 4424.0, 2631.5, 4431.0, 2639.5, 4455.0, 2643.5, 4470.0, 2643.5, 4508.0, 2630.5, 4510.5, 2627.0, 4511.5, 2617.0, 4516.5, 2603.0, 4539.5, 2552.0, 4523.5, 2530.0, 4501.0, 2512.5, 4454.0, 2491.5, 4420.0, 2471.5, 4399.5, 2449.0, 4388.5, 2427.0, 4375.5, 2394.0, 4362.5, 2368.0, 4352.0, 2360.5, 4348.5, 2355.0, 4343.5, 2324.0, 4355.5, 2302.0, 4370.5, 2280.0, 4363.0, 2251.5, 4343.5, 2256.0, 4337.5, 2282.0, 4322.5, 2312.0, 4298.5, 2340.0, 4323.5, 2445.0, 4323.0, 2447.5, 4316.0, 2449.5, 3752.0, 2482.5, 3726.0, 2481.5, 3725.5, 2475.0, 3729.5, 2470.0, 3795.5, 2398.0, 3817.5, 2370.0, 3819.5, 2347.0, 3819.5, 2338.0, 3817.5, 2331.0, 3808.0, 2324.5, 3786.0, 2319.5, 3731.5, 2383.0, 3698.5, 2411.0, 3692.5, 2460.0, 3699.5, 2481.0, 3708.5, 2486.0, 3717.5, 2504.0, 3715.5, 2533.0, 3717.5, 2566.0, 3715.5, 2579.0, 3714.5, 2614.0, 3710.5, 2630.0, 3669.5, 2660.0, 3658.0, 2684.5, 3627.0, 2707.5, 3589.0, 2727.5, 3548.0, 2758.5, 3481.0, 2758.5, 3381.0, 2704.5, 3343.5, 2657.0, 3318.0, 2632.5, 3296.0, 2617.5, 3251.0, 2623.5, 3196.0, 2619.5, 3161.0, 2606.5, 3119.0, 2583.5, 3094.5, 2547.0, 3051.5, 2510.0, 3027.5, 2478.0, 3009.5, 2442.0, 3000.5, 2408.0, 2977.5, 2379.0, 2972.5, 2333.0, 2970.5, 2262.0, 2973.5, 2150.0, 2996.0, 2127.5, 3021.5, 2108.0, 3037.5, 2080.0, 3045.5, 2059.0, 3058.0, 2039.5, 3095.0, 2021.5, 3116.0, 2004.5, 3139.0, 1963.5, 3204.5, 1931.0, 3200.5, 1922.0, 3185.5, 1910.0, 3192.5, 1882.0, 3203.5, 1859.0, 3208.5, 1853.0, 3241.5, 1817.0, 3266.0, 1795.5, 3281.5, 1779.0, 3284.5, 1748.0, 3294.5, 1730.0, 3336.0, 1710.5, 3351.0, 1706.5, 3392.0, 1683.5, 3450.0, 1671.5, 3448.5, 1651.0, 3455.5, 1631.0, 3480.0, 1618.5, 3561.0, 1609.5, 3584.0, 1595.5, 3642.0, 1595.5, 3665.5, 1593.0, 3663.0, 1574.5, 3652.0, 1574.5, 3632.0, 1580.5, 3612.0, 1570.5, 3584.5, 1543.0, 3564.5, 1502.0, 3545.5, 1477.0, 3540.5, 1465.0, 3519.5, 1403.0, 3499.5, 1350.0, 3479.5, 1286.0, 3475.5, 1208.0, 3482.5, 1080.0, 3518.5, 940.0, 3556.5, 887.0, 3617.5, 816.0, 3699.0, 751.5, 3815.0, 697.5, 3909.0, 669.5, 4044.0, 664.5, 4087.0, 656.5, 4171.0, 664.5, 4261.0, 682.5, 4266.0, 684.5, 4354.0, 748.5, 4426.5, 823.0, 4504.5, 1012.0, 4522.5, 1098.0, 4517.5, 1195.0, 4480.5, 1320.0, 4391.5, 1451.0, 4354.5, 1513.0, 4347.5, 1531.0, 4305.5, 1662.0, 4328.0, 1669.5, 4390.0, 1682.5, 4433.0, 1697.5, 4477.0, 1725.5, 4500.0, 1754.5, 4524.5, 1772.0, 4538.5, 1851.0, 4574.0, 1871.5, 4661.5, 1963.0, 4672.5, 1975.0, 4701.5, 2025.0, 4728.0, 2035.5, 4764.0, 2059.5, 4820.0, 2100.5, 4858.5, 2147.0, 4886.0, 2187.5, 4928.0, 2208.5, 4930.5, 2211.0, 4972.0, 2273.5, 4996.0, 2279.5, 5020.0, 2282.5, 5102.0, 2338.5, 5156.5, 2406.0, 5183.5, 2459.0, 5187.5, 2496.0, 5157.5, 2543.0, 5145.5, 2587.0, 5133.5, 2603.0, 5124.5, 2628.0, 5122.0, 2631.5, 5090.0, 2648.5, 5077.0, 2653.5, 5058.0, 2686.5, 5037.0, 2700.5, 5011.0, 2724.5, 4989.0, 2741.5, 4983.0, 2744.5, 4947.0, 2751.5, 4919.5, 2792.0, 4853.0, 2868.5, 4795.0, 2895.5, 4771.0, 2902.5, 4737.0, 2909.5, 4709.0, 2903.5, 4691.0, 2914.5]], 'width': 5456, 'height': 3632}, {'id': 8216, 'image_id': 3452, 'category_id': 2, 'iscrowd': 0, 'area': 2590712, 'bbox': [2968.0, 658.0, 2218.0, 2256.0], 'segmentation': [[4669.0, 2913.5, 4641.0, 2909.5, 4588.0, 2909.5, 4553.0, 2901.5, 4527.5, 2874.0, 4510.0, 2848.5, 4498.0, 2839.5, 4470.0, 2831.5, 4459.5, 2812.0, 4439.5, 2800.0, 4437.5, 2785.0, 4437.5, 2728.0, 4419.5, 2718.0, 4400.5, 2677.0, 4390.5, 2663.0, 4402.0, 2642.5, 4424.0, 2631.5, 4436.0, 2644.5, 4439.0, 2644.5, 4469.0, 2639.5, 4512.0, 2628.5, 4517.5, 2617.0, 4524.5, 2582.0, 4543.5, 2558.0, 4532.0, 2539.5, 4507.0, 2528.5, 4481.0, 2508.5, 4432.0, 2484.5, 4412.5, 2462.0, 4400.5, 2440.0, 4381.5, 2414.0, 4370.5, 2373.0, 4368.0, 2368.5, 4348.5, 2355.0, 4343.5, 2336.0, 4344.5, 2316.0, 4354.5, 2298.0, 4359.5, 2285.0, 4362.5, 2263.0, 4362.5, 2259.0, 4360.0, 2255.5, 4342.5, 2254.0, 4344.5, 2264.0, 4334.5, 2300.0, 4311.5, 2322.0, 4303.5, 2338.0, 4322.5, 2443.0, 4322.0, 2445.5, 4314.0, 2446.5, 3715.0, 2484.5, 3718.5, 2479.0, 3789.5, 2400.0, 3804.5, 2376.0, 3817.5, 2359.0, 3821.5, 2346.0, 3811.0, 2334.5, 3790.0, 2319.5, 3720.0, 2385.5, 3696.5, 2412.0, 3697.5, 2437.0, 3695.5, 2477.0, 3702.5, 2491.0, 3719.5, 2510.0, 3719.5, 2538.0, 3714.5, 2547.0, 3719.5, 2558.0, 3719.5, 2562.0, 3715.5, 2581.0, 3711.5, 2637.0, 3669.0, 2661.5, 3649.0, 2698.5, 3631.0, 2707.5, 3603.0, 2717.5, 3592.0, 2727.5, 3560.0, 2747.5, 3524.0, 2761.5, 3487.0, 2753.5, 3456.0, 2752.5, 3402.0, 2722.5, 3369.5, 2697.0, 3343.5, 2641.0, 3332.5, 2626.0, 3308.0, 2612.5, 3287.0, 2613.5, 3252.0, 2619.5, 3206.0, 2624.5, 3200.0, 2624.5, 3165.0, 2617.5, 3127.0, 2581.5, 3099.0, 2560.5, 3079.0, 2550.5, 3056.0, 2536.5, 3032.5, 2495.0, 3010.5, 2472.0, 2999.5, 2420.0, 2972.5, 2340.0, 2972.5, 2303.0, 2967.5, 2229.0, 2976.5, 2154.0, 2983.5, 2140.0, 3012.5, 2106.0, 3017.0, 2101.5, 3041.5, 2089.0, 3039.5, 2080.0, 3045.5, 2069.0, 3050.5, 2048.0, 3069.0, 2028.5, 3105.0, 2020.5, 3129.0, 1978.5, 3157.0, 1960.5, 3179.0, 1943.5, 3206.5, 1928.0, 3199.5, 1923.0, 3191.5, 1913.0, 3186.5, 1894.0, 3199.5, 1877.0, 3206.5, 1862.0, 3235.5, 1824.0, 3272.5, 1787.0, 3275.5, 1770.0, 3289.0, 1739.5, 3305.0, 1729.5, 3344.0, 1711.5, 3360.0, 1698.5, 3371.0, 1692.5, 3399.0, 1680.5, 3436.0, 1669.5, 3444.0, 1669.5, 3461.0, 1621.5, 3509.0, 1612.5, 3551.0, 1611.5, 3569.0, 1606.5, 3586.0, 1596.5, 3661.5, 1593.0, 3659.0, 1568.5, 3632.0, 1583.5, 3605.5, 1566.0, 3577.5, 1524.0, 3552.5, 1491.0, 3508.5, 1387.0, 3492.5, 1355.0, 3476.5, 1278.0, 3476.5, 1218.0, 3473.5, 1161.0, 3502.5, 979.0, 3504.5, 973.0, 3510.5, 964.0, 3599.0, 835.5, 3786.0, 705.5, 3867.0, 673.5, 4048.0, 662.5, 4088.0, 661.5, 4125.0, 657.5, 4153.0, 660.5, 4187.0, 673.5, 4287.0, 695.5, 4317.0, 715.5, 4346.0, 739.5, 4421.5, 810.0, 4466.5, 885.0, 4496.5, 941.0, 4516.5, 985.0, 4529.5, 1155.0, 4505.5, 1248.0, 4492.5, 1323.0, 4475.0, 1333.5, 4465.0, 1338.5, 4455.0, 1340.5, 4452.5, 1343.0, 4402.5, 1444.0, 4385.5, 1462.0, 4371.5, 1503.0, 4352.5, 1527.0, 4342.5, 1568.0, 4298.5, 1669.0, 4333.0, 1669.5, 4374.0, 1672.5, 4408.0, 1684.5, 4448.0, 1704.5, 4500.0, 1741.5, 4544.5, 1788.0, 4545.5, 1847.0, 4547.5, 1850.0, 4564.0, 1870.5, 4609.5, 1897.0, 4710.5, 2035.0, 4780.0, 2068.5, 4832.0, 2110.5, 4835.5, 2114.0, 4883.5, 2196.0, 4976.0, 2268.5, 4979.0, 2270.5, 5025.0, 2274.5, 5103.0, 2328.5, 5146.5, 2378.0, 5165.5, 2427.0, 5185.5, 2464.0, 5179.5, 2512.0, 5175.0, 2518.5, 5159.5, 2531.0, 5155.5, 2571.0, 5153.5, 2578.0, 5139.5, 2600.0, 5092.5, 2649.0, 5054.0, 2692.5, 4998.0, 2735.5, 4946.0, 2756.5, 4931.5, 2775.0, 4913.5, 2803.0, 4875.0, 2839.5, 4839.0, 2866.5, 4802.0, 2882.5, 4777.0, 2895.5, 4746.0, 2905.5, 4695.0, 2905.5, 4669.0, 2913.5]], 'width': 5456, 'height': 3632}]}
Traceback (most recent call last):
  File "/root/anaconda3/lib/python3.6/site-packages/mask_rcnn-2.1-py3.6.egg/mrcnn/model.py", line 1695, in data_generator
    use_mini_mask=config.USE_MINI_MASK)
  File "/root/anaconda3/lib/python3.6/site-packages/mask_rcnn-2.1-py3.6.egg/mrcnn/model.py", line 1209, in load_image_gt
    image = dataset.load_image(image_id)
  File "/root/anaconda3/lib/python3.6/site-packages/mask_rcnn-2.1-py3.6.egg/mrcnn/utils.py", line 367, in load_image
    image = skimage.color.gray2rgb(image)
  File "/root/anaconda3/lib/python3.6/site-packages/skimage/color/colorconv.py", line 862, in gray2rgb
    raise ValueError("Input image expected to be RGB, RGBA or gray.")
ValueError: Input image expected to be RGB, RGBA or gray.

TypeError: Expected bytes, got list

Hey @waspinator , thanks for this work. I have a problem about the decoding of RLE.
I set 'is_crowd' to 1 to make RLE format in shapes_to_coco.py. But I use MaskRCNN from facebookresearch and have a bug related to your program.

File "C:\Users\DTMLLUAdminUser\Anaconda3\envs\detectron2\lib\site-packages\pycocotools\mask.py", line 91, in decode
return _mask.decode([rleObjs])[:,:,0]
File "pycocotools_mask.pyx", line 146, in pycocotools._mask.decode
File "pycocotools_mask.pyx", line 128, in pycocotools._mask._frString
TypeError: Expected bytes, got list

do you have any idea or what should I do to finish this bug. Thank you.

How do I cite this repository?

Hey @waspinator, thanks for this work. I have used this in my master's thesis and I would like to know how I should cite this repository. Thank you.

Are you planning to add the keypoint

Please try to add the keypoint to your code for the bodypose.
Thank you

Bug: mismatch of "bbox" and "segmentation"

Dictionary below is first images annotations information that made by pycococreator.
This link is an image what I use for explain my question.

https://light-tree.tistory.com/123

As I know, "bbox" contains left top coordinate and right bottom coordinate.

To simple check data is fine, I printed max value of "bbox" and "segmentation".
As I understand, max value of "segmentation" and max value "bbox" have to be same.
(whatever that max value things means right bottom x or y)

but, max of "bbox" is 765 that right bottom y coordinate
and max of "segmentation" is 787.5 that point is on shoe bottom in my image.

Is this my mistake? How can I fix this?

{"bbox": [93.0, 23.0, 254.0, 765.0], "segmentation": [[320.0, 787.5, 284.0, 787.5, 260.0, 781.5, 233.0, 779.5, 227.5, 771.0, 228.5, 762.0, 233.5, 756.0, 231.0, 742.5, 227.0, 747.5, 220.0, 749.5, 210.0, 749.5, 201.0, 743.5, 196.0, 748.5, 181.0, 748.5, 169.0, 744.5, 148.5, 725.0, 141.5, 707.0, 132.0, 705.5, 122.0, 699.5, 107.5, 680.0, 107.5, 671.0, 115.0, 661.5, 136.0, 648.5, 152.0, 645.5, 153.5, 638.0, 164.0, 631.5, 176.0, 616.5, 204.5, 593.0, 206.5, 588.0, 203.0, 571.5, 196.0, 578.5, 187.5, 575.0, 187.5, 571.0, 195.5, 562.0, 195.5, 557.0, 186.0, 545.5, 157.0, 549.5, 149.5, 542.0, 151.5, 528.0, 144.5, 510.0, 144.5, 493.0, 132.5, 455.0, 129.5, 418.0, 132.5, 374.0, 118.5, 316.0, 109.5, 298.0, 103.5, 267.0, 97.5, 256.0, 92.5, 235.0, 93.5, 218.0, 110.5, 188.0, 136.0, 163.5, 144.5, 161.0, 140.0, 159.5, 133.5, 150.0, 134.5, 129.0, 127.0, 126.5, 112.5, 108.0, 112.5, 101.0, 116.5, 95.0, 115.5, 75.0, 119.5, 70.0, 119.5, 59.0, 146.0, 30.5, 166.0, 22.5, 190.0, 23.5, 215.0, 30.5, 246.5, 65.0, 247.5, 107.0, 253.5, 124.0, 250.5, 159.0, 244.0, 166.5, 226.0, 173.5, 221.5, 181.0, 232.0, 183.5, 245.5, 193.0, 272.5, 237.0, 291.5, 257.0, 299.5, 270.0, 303.5, 284.0, 303.5, 304.0, 299.5, 324.0, 291.5, 341.0, 313.5, 397.0, 324.5, 416.0, 335.5, 428.0, 333.5, 443.0, 322.5, 455.0, 330.5, 491.0, 338.5, 508.0, 337.5, 516.0, 333.0, 519.5, 324.0, 519.5, 320.5, 529.0, 321.5, 539.0, 330.5, 555.0, 331.5, 571.0, 323.5, 583.0, 297.5, 605.0, 292.0, 617.5, 270.5, 621.0, 269.5, 634.0, 274.5, 651.0, 269.5, 657.0, 267.5, 672.0, 268.5, 693.0, 280.5, 727.0, 288.0, 729.5, 296.0, 741.5, 303.0, 741.5, 309.0, 736.5, 317.0, 737.5, 333.0, 755.5, 341.0, 756.5, 346.5, 763.0, 346.5, 768.0, 339.0, 778.5, 320.0, 787.5]], "area": 114927, "id": 1, "category_id": 1, "height": 1080, "image_id": 1, "iscrowd": 0, "width": 1920}

Thanks.

The installed module is not found in Jupyter Notebook

Thanks for sharing your library.
But your module 'pycocotools' is not found in your Jupyter Notebook example 'visualize_coco'.
after installation.
I can find a library named 'pycococreatortools' by library searching using 'pip list' command
but not pycocotools.

Is there any specific requirement for your library?
I currently use Anaconda 3 environment and I installed it using Anaconda Prompt.

I am completely new to python language.

Question: overlapping annotations?

Does this happen to support overlapping masks? For example 'car' and 'tire'. Also I can see how that would be represented as different objects in an XML file, but how would that be represented in a jpeg mask? Thanks!

I see that there's the parameter is_crowd, but not sure how that may be determined from a jpeg mask.

different objects of the same class in a black and white mask diagram

Mask RCNN and RLE

Hello everyone,

I am running the matterport Mask_RCNN on my own dataset created using waspinator/pycococreator

Everything works fine when I am using polygons, but as soon as I am using RLE because I have crowded images I got an error when trying to display a few images from the training dataset,

I got problems with images with holes using polygons and I assumed that is why the pycococreator uses RLE when the image is crowded. This part works perfectly, but then since MaskRCCN requires polygons I am stuck in this loop.

Is there a way to convert RLE to Polygons? I am not even sure this would solve the problem

result from shapes_to_coco.py is strange

Hello.
I run the shapes_to_coco.py code using the shape dataset, but the result is strange.

I wonder if you can guess where this problem is coming from. Maybe there is a problem in converting the mask.