davidbrai / deep-learning-traffic-lights Goto Github PK

View Code? Open in Web Editor NEW

478.0 35.0 161.0 14.07 MB

Code and files of the deep learning model used to win the Nexar Traffic Light Recognition challenge

Home Page: https://medium.com/@davidbrai/recognizing-traffic-lights-with-deep-learning-23dae23287cc

License: BSD 2-Clause "Simplified" License

Jupyter Notebook 99.94% Python 0.06%

deep-learning caffe recognize-traffic-lights

deep-learning-traffic-lights's Introduction

Recognizing Traffic Lights with Deep Learning

This repo contains the files used to train and run the classifier described in this blog post. This was done during a challenge by Nexar to recognize traffic lights based on images taken by their dashcam app.

Dependencies

Caffe with python bindings.

Directory contents:

/model: contain a caffe deploy.prototxt file and three weights files. The three weights files are used together in a model ensemble.

/testing: has jupyter notebook files that run the model and perform the weighted average.

/training: contains the files needed to train the model (except the training data)

Training the model

The images were first converted to lmdb format and resized to 256x256 using this command:

GLOG_logtostderr=1 ~/caffe/build/tools/convert_imageset \
    --resize_height=256 --resize_width=256 --shuffle  \
    ~/nexar/images/ \
    ~/nexar/labels_test.txt \
    ~/nexar/lmdb/test_lmdb

Each model has a directory in training with some or all of the following files:

solver.prototxt   caffe solver file
solver_p2.prototxt  caffe solver file with lower base learning rate
train_val.prototxt  network training file
rotation_layer.py   python caffe layer for data augmentation with rotation

squeeze_net_manual_scratch__os was training from scratch. The other two models were fine-tuning from weights trained on ImageNet. The weights file is named squeezenet_v1.0.caffemodel.

deep-learning-traffic-lights's People

Contributors

Stargazers

Watchers

Forkers

ossdc klonikar dassiorleando rahulmirdha jinz1 allensmile hooin pyous chagge caomw benjamesbabala ieyer anazou akjagadish 4skynet llpj soledad89 sergeytimoshin mehrdad-shokri jianghobryan fototo aihgf ningzhou shirleyyim lxj0276 xclu haroldss szm88 stella-gao rcholic kjeanclaude tpjg arasharchor sg47 aloshkad kenburke horizonhao gnahz-eh collector-m anuragreddygv323 hanuor leempan superpoca tonychouzju tairycy zaxon zoujg xujuntwt95329 jeanphilippemonteuuis lyk125 leezqcst dreadlord1984 deepanonymous fenix0817 caplu zsxgb yangsuo zepzhang suzhenghang crucified713 rap9430 nature0310 zhanglamei vtpp2014 stgitorious bandokal wuyuzaizai cjjsaic yehudaitkin sagar72 vinibc darwinbeing mtmzn jxlin lostinet abigriver chengsonghust chenming5828 codermckee 521314 siam7tong ninopereira snowbhr06 sherlockls leooo48 svmtracking carabob robbiewu008 sanskar107 sheuan yunhaoqiao zhuxiang555111 yangkuiking tdautc19841202 mariankh yaoling13 swwwu dingranshiwo bikmaeff whrenstone

deep-learning-traffic-lights's Issues

Dataset

Great project. I'm trying to give it a go. Do you think you can make available the dataset?

I want to get the data

Hello,can you give me the data of the traffic light,I can not download the data from Nexar.

sliding_patch.ipynb

when I use sliding_patch.ipynb

from caffe.classifier import Classifier

model_def = "/home/zsx/deep-learning-traffic-lights-master/model/deploy.prototxt"
model_weights = "/home/zsx/deep-learning-traffic-lights-master/model/train_squeezenet_trainval_manual_p2__iter_3817.caffemodel"

c = Classifier(
model_def,
model_weights,
mean=np.array([104, 117, 123]),
raw_scale=255,
channel_swap=(2,1,0),
image_dims=(256, 256))

set batch size

BATCH_SIZE = 64
c.blobs['data'].reshape(BATCH_SIZE, 3, c.blobs['data'].shape[2], c.blobs['data'].shape[3])
c.blobs['prob'].reshape(BATCH_SIZE, 3)
c.reshape()

def class_idx_to_name(idx):
return ['none', 'red', 'green'][idx]

[I 21:58:49.510 NotebookApp] KernelRestarter: restarting kernel (1/5)
WARNING:root:kernel 09a50faf-6028-42f2-8e9b-3294050616e8 restarted
My ipython on the restart, which is why?

can i get your dataset for learn？

I'll be greatful if you contact me. [email protected]

AttributeError: 'dict' object has no attribute 'itervalues'

hello, I use python3.5 to run test demo edited with notepad++, and it leads to the error "AttributeError: 'dict' object has no attribute 'itervalues'" as the following, how can i solve it? Your help will be greatly appreciated.
————————————————————————————————————————————
The error is fixed with changing my python path… you can close this Issue， thx.

About the Dataset

Hi David,
I'm very interested in your work, and I want to do some experiments about it too, but the Nexar Challenge #1 has been closed. Could you please send the challenge dataset to me. Thank you very much.

Asking for the Dataset

Hi. May I have the dataset of the NEXAR Challenge One? The challenge has been closed and I cannot download the dataset from the website. Thank you for your time.

how to use notebook in aws

Hello, I used AWS service. There is a problem, would like to ask you again

In my AWS use I used a P2 xlarge, this in my use, aws service staff told me that the use of ssh link to use AWS p2.

There is no way to use jupyter notebook. Do you use the desktop UI? How to use the browser-based jupyter notebook.

If you can be convenient, give me a little share

Thanks,
Madhu.