Programmerwyl

@GustavZ Thank you very much for your reply. yes ,it is the current master I haven't changed anything about your code. python 2.7 opencv 4.0 jetpack 28.2

I successfully run it on video， I use 2 different frameworks (Darknet and Keras). Just extract the method of loading the model to the outermost and it works

1.Load all models in the data directory 2.Read the video in a loop 3.For a frame of video，run vehicle_detection.py,license_plate_detection.py,license_plate-_ocr.py one by one note:The detect function of darknet only accepts file...

This is an annotation tool If there is not label info in the .txt corresponding to the image, NO LABEL will be displayed This indicates that you need to enter...

For the label box of the license plate, it is also marked by the operation area of these keys, such as c, a, d, s To be honest, it doesn't...

当训练集比较小的时候，尤其是重新训练的话，如果以作者原先的lr,确实容易出现acc一直为0的情况 我的训练样本700个，测试样本300个，lr初始值 0.0005，训练时acc一直为0 当lr提高10倍，lr初始值0.005时，一开始acc也是0，不过随着训练次数的增加，acc不再为0 如果训练不是从头开始，而是基于已经训练过的模型，则能够缩短acc为0的时间 还有就是注意观察loss，如果loss大于10，则很有可能acc为0 当loss小于8的时候，acc不再为0 D:\develop\Anaconda3\envs\alpr-unconstrained-master\python.exe D:/data/project/CTPN_CTC/train/train_lp.py Using TensorFlow backend. 2019-04-22 16:06:40.011247: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX...

提高lr是为了提高学习速度，而增大step是为了延长训练的时间，他们都是可以解决acc为0的情况的

Which model is used to train 1k3d68.onnx? How do I convert a trained model into onnx?