In this project, we implemented a recognition and a classification model. The recognition system was used the darknet framework and the YOLO v3 model to detect traffic signs in a frame of a video. Whereas, the classification system was based on convolutional neural network (CNN). The YOLO model determined where the traffic sign was located in the frame and the CNN model determined the class of that traffic sign. We used a dataset with 21 classes of Traffic signs.
- Speed Limit - 20
- Speed Limit - 30
- No Entry
- Right Turn Prohibited
- Left Turn Prohibited
- Horn Prohibited
- No Parking
- No Stopping or Standing
- Compulsory Ahead
- Compulsory Left Ahead
- Straight or Right
- Speed Limit: 50
- Compulsory Right Ahead
- Speed Limit: 60
- Speed Limit: 70
- Speed Limit: 90
- Speed Limit: 100
- Speed Limit: 120
- Stop
- Give Way
The results generated from the model can be viewed here.