search

hellojialee / Traffic_Sign_Recognition_Efficient_CNNs

A repository for the paper "Real-Time Traffic Sign Recognition Based on Efficient CNNs in the Wild"
MIT License
139 stars 56 forks source link

CPU-Mode #8

Closed langelesl closed 5 years ago

langelesl commented 5 years ago

Hello Jia. Thank you for sharing your wonderful codes. I noticed you've uploaded your weight and test_frcnn.py for testing in the detection part. Is it possible to test your code in CPU-Mode without CUDA? If positive, could you please share your tips about modifying your code? I will appreciate your help.

hellojialee commented 5 years ago

You can just change os.environ["CUDA_VISIBLE_DEVICES"] = "0" # use GTX 1080 to os.environ["CUDA_VISIBLE_DEVICES"] = " " # use GTX 1080 in https://github.com/USTClj/Traffic_Sign_Recognition_Efficient_CNNs/blob/2998d56e45d625c1b35a31ff28467aaf092ec627/Detection/mobilenet_frcnn_detector/test_frcnn.py#L17

Our final inner code is actually follows Mask_RCNN.

langelesl commented 5 years ago

@USTClj Great, I will try it later. Thanks for your reply.