senofsky / face-mask-detector Goto Github PK

A library and command line utility for detecting face masks in images and video streams

License: MIT License

Makefile 5.40% Python 94.60%

computer-vision covid-19 deep-learning detect-face-masks face-mask-detection face-mask-detector face-masks keras keras-tensorflow machine-learning mobilenetv2 python python3 video-stream

face-mask-detector's People

Contributors

Stargazers

Watchers

Forkers

kareemsasa3 cemot bharathivasan informaticacba

face-mask-detector's Issues

Add logging

face-mask-detector should have logging capabilities.

I think command_line_interface.py should configure the logger with a given log level (via the verbose argument) and the rest of the library should use the logger.

Model trainer switches labels

train_face_mask_detector.py creates a model that "switches" the labels, so:

a positive face mask detection is labeled as negative
a negative face mask detection is labeled as positiive

Declare supported python versions

Add a badge of the supported python versions

Documenting which python versions we support is helpful to other developers.

We should document the supported python versions and add them as badges in the README.

Add docs

Add documentation on how the models work and how to use the library.

Expand training data to include people wearing patterned face masks

The current face_mask_detector.model struggles at detecting patterned face masks.

The current dataset only contains images of more "standard" masks. In order to improve the model's accuracy, we need to expand the dataset to include people wearing patterned face masks.

Fix type errors caused by external libraries

The following is a snippet of received type errors:

face_mask_detector/lib.py:4: error: Cannot find implementation or library stub for module named 'cv2'
face_mask_detector/lib.py:6: error: Cannot find implementation or library stub for module named 'imutils'
face_mask_detector/lib.py:7: error: Skipping analyzing 'numpy': found module but no type hints or library stubs
face_mask_detector/lib.py:11: error: Cannot find implementation or library stub for module named 'imutils.video'
face_mask_detector/lib.py:12: error: Cannot find implementation or library stub for module named 'tensorflow.keras.applications.mobilenet_v2'
face_mask_detector/lib.py:13: error: Cannot find implementation or library stub for module named 'tensorflow.keras.preprocessing.image'
face_mask_detector/lib.py:14: error: Cannot find implementation or library stub for module named 'tensorflow.keras.models'

This resource may be of use.

Fix typing errors in command_line_interface.py

Currently, mypy is not passing due to a few typing errors.

Dataset ...

Hi,
Thank you for sharing this excellent project.
I need to improve the model by increasing the number of pictures (adding more real images) to the data set. My questions are:

In order to improve the model, do I need just to copy images (with/without face mask) and paste it in corresponding folders or do I need to do extra steps?
If I need to do something else, please direct me to do so.

thank you very much

Refactor box-related logic into Box class

The box-related logic is duplicated throughout lib.py and would be fixed if it was consolidated into a Box class.

command_line_interface.py should accept video sources as arguments

Currently, command_line_interface.py only accepts images as arguments.

It should also accept different video sources instead of only supporting the webcam.

Fix typing errors in lib.py

Currently, mypy is not passing due to a few typing errors.

Fix warnings in train_face_mask_detector.py

python scripts/train_face_mask_detector.py -d dataset will print the following warnings during execution:

/home/senofsky/projects/face-mask-detector/venv/lib/python3.8/site-packages/tensorflow/python/ops/random_ops.py:285: SyntaxWarning: "is" with a literal. Did you mean "=="?  minval_is_zero = minval is 0  # pylint: disable=literal-comparison
/home/senofsky/projects/face-mask-detector/venv/lib/python3.8/site-packages/tensorflow/python/ops/random_ops.py:286: SyntaxWarning: "is" with a literal. Did you mean "=="?  maxval_is_one = maxval is 1  # pylint: disable=literal-comparison
/home/senofsky/projects/face-mask-detector/venv/lib/python3.8/site-packages/tensorflow/python/ops/ragged/ragged_batch_gather_with_default_op.py:84: SyntaxWarning: "is not" with a literal. Did you mean "!="?  if (default_value.shape.ndims is not 0
/home/senofsky/projects/face-mask-detector/venv/lib/python3.8/site-packages/tensorflow/python/ops/ragged/ragged_batch_gather_with_default_op.py:85: SyntaxWarning: "is not" with a literal. Did you mean "!="?  and default_value.shape.ndims is not 1):
WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.WARNING:`input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.
2020-07-16 19:26:24.988178: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2020-07-16 19:26:24.988200: E tensorflow/stream_executor/cuda/cuda_driver.cc:313] failed call to cuInit: UNKNOWN ERROR (303)
2020-07-16 19:26:24.988219: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (spirit-of-fire): /proc/driver/nvidia/version does not exist2020-07-16 19:26:24.988390: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2020-07-16 19:26:25.009288: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 3001000000 Hz
2020-07-16 19:26:25.009793: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f4a2c000b60 initialized for platform Host (this does not guarantee that XLA will be used). Devices:2020-07-16 19:26:25.009812: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version

Add version badge to README

We should add a badge indicating the version of face-mask-detector to the README so it can be quickly identified.

Ideally, this should be the first badge to be displayed.

Improve error messages for invalid arguments

Currently, _validate_args() will log an error stating that a directory is not writeable if it does not exist.

While this is technically true, we should notify the users why the directory isn't writeable; notifying the user that the directory does not exist or the permissions are set to read-only is more helpful.

We should always strive to provide a positive user experience.