jacobssy/traffic_sign_detection

Traffic_Sign_detection

完成交通标志的识别分类（基于GTSBR数据），测试数据集上达到98.84%的准确率

检测部分的特征提取层包含YoloV2,yoloV2Tiny,yoloV3,MobileNet,SqueezeNet,ShuffleNet,And a Structure deSign by me:S-MobileNet

分类和检测的训练测试详情见:classify/readme.md detection/readme.md

需要weights详见：#4

--- 2018.06.15本科毕设答辩结束，该repo不再更新（也许还会再去做一下，毕竟这个repo确实比较dirty）

traffic_sign_detection's Issues

Model weights?

Great project! I was wondering if you could attach the model weights you trained?

Problem in input images

Hello @jacobssy again , Buddy can you tell how to resolve this error? The GTSRB has 2 folders final_test & final_training & in this they have .ppm file. I assume it has error in this ; please help me to resolve the issue.
thanking you
sayantan.

version of keras

Hi,

I tried to run predict.py in detection folder with either keras 1 or keras 2, but none would work due to import problems.

What's the exact version of Keras are you using?

Thanks!

Slowing In epoch step

Hello, @jacobssy again buddy i successfully reached in input Epoch steps. It is running very slowly i think GPU is not working in the code which i validated with Task manager which is using CPU & GPU is not been used. I think that your computer has good specification. What are your comp specifications.
Help would be apprecialted.
THANK YOU again.

Issue with file in YOLO step by step (forward)

Buddy @jacobssy hello again are you using python 3 , I will ask you doubts if I encounter any problem 1st problem - in which step you get the weights_traffic.h5 in you input? In your code
Thank you
Regards Sayantan