1. Setup dataset

   $ wget https://sid.erda.dk/public/archives/daaeac0d7ce1152aea9b61d9f1e19370/GTSRB_Final_Training_Images.zip
   $ unzip GTSRB_Final_Training_Images.zip
   $ wget https://sid.erda.dk/public/archives/daaeac0d7ce1152aea9b61d9f1e19370/GTSRB_Final_Test_Images.zip
   $ unzip GTSRB_Final_Test_Images.zip
   $ wget https://sid.erda.dk/public/archives/daaeac0d7ce1152aea9b61d9f1e19370/GTSRB_Final_Test_GT.zip
   $ unzip GTSRB_Final_Test_GT.zip
   $ mv GT-final_test.csv GTSRB/Final_Test/Images

2. Data augmentation and generate train/test pickle file

   $ python gen_pickle.py

3. Training

   $ python train.py
   [epoch 0] Mini-batch loss at 0: 491.711456
   [epoch 0] Minibatch accuracy: 4.7%
   ...
   [epoch 1] Mini-batch loss at 63488: 24.210030
   [epoch 1] Minibatch accuracy: 93.8%
   Test accuracy: 81.5%
   Model saved in file: models/deep_traffic_sign_model

4. Detection

   $ python detect_traffic_sign.py [target_image_file]
   

MIT