In this project, I have implemented a classifier using deep neural networks, convolutional neural networks and transfer learning to classify traffic signs. I have trained a model so that it can decode traffic signs from natural images by using the German Traffic Sign Dataset. After the model is trained, I have tested my model program on new images of traffic signs collected from the web and captured images of traffic sign locally collected from a friend.
- Download the dataset. This is a pickled dataset in which we've already resized the images to 32x32.
#Test Results
Follow the keras version traffic-sign-classification-with-keras.ipynb for updated results
###Validation accuracy 99.31 %
###Test accuracy 95.44 %
#Model architecture
| Layer (type) | Output Shape | Param # | Connected to |
|---|---|---|---|
| convolution2d_11 (Convolution2D) | (None, 28, 28, 6) | 456 | convolution2d_input_9[0][0] |
| maxpooling2d_10 (MaxPooling2D) | (None, 14, 14, 6) | 0 | convolution2d_11[0][0] |
| activation_10 (Activation) | (None, 14, 14, 6) | 0 | maxpooling2d_10[0][0] |
| convolution2d_12 (Convolution2D) | (None, 10, 10, 16) | 2416 | activation_10[0][0] |
| maxpooling2d_11 (MaxPooling2D) | (None, 5, 5, 16) | 0 | convolution2d_12[0][0] |
| activation_11 (Activation) | (None, 5, 5, 16) | 0 | maxpooling2d_11[0][0] |
| flatten_7 (Flatten) | (None, 400) | 0 | activation_11[0][0] |
| dense_15 (Dense) | (None, 128) | 51328 | flatten_7[0][0] |
| dropout_11 (Dropout) | (None, 128) | 0 | dense_15[0][0] |
| activation_12 (Activation) | (None, 128) | 0 | dropout_11[0][0] |
| dense_16 (Dense) | (None, 43) | 5547 | activation_12[0][0] |
| Total params: 59747 |
Trained on 26270 samples, validated on 12939 samples