ND-recognition-of-traffic-signs

2019ML-Detection and recognition of traffic signs for self-driving applications

instruction

to run this, you should:

1. install and learn Tensorflow object detection API.

2. download GTSDB and GTSRB dataset.

3. copy visualization_utils.py to C:*\models\research\object_detection\utils

4. detection:

first,download faster_rcnn_resnet_101 model. then copy it to ND-recognition-of-traffic-signs/traffic-sign-detection-master/models(you should new this folder by yourself) , download list: https://drive.google.com/open?id=15OxyPlqyOOlUdsbUmdrexKLpHy1l5tP9 ND second, run ND-recognition-of-traffic-signs/traffic-sign-detection-master/Run_models_on_new_images-Copy1.ipynb (you should edit the path in that file)

5. classification:

from the step4 we get the output img in ND-recognition-of-traffic-signs/traffic-sign-detection-master/output. then run classification.ipynb to get the result.

what should do next:

1. Find a way to test the model in engineering and evaluate the accuracy.

2. Change the model, compare the differences, find the most appropriate one.

to get other trained models, download in here:

https://github.com/aarcosg/traffic-sign-detection

by Lei, 2019.11.24