artyze / yolov3_lite Goto Github PK

Dear author, i just want to train the model by quantization in the process. How to convert the google quantization model to the original format

do you have the code for compress the yolo-v3 model, as yolo-v3 model with 416&416 need GPU memory 2~3GB, so do you have any suggest for reduce the memory?

author ,
thanks for your providision, i want to run google-quant of this repo.
but idont konw how to get output_scale and output_zero_point in the cfg file.
output_scale = 0, 0, 0.0875, 0, 0.075846, 0, 0.069873, 0, 0.050603, 0, 0.031124, 0, 0.023692, 0.021359, 0.041686
output_zero_point = 0, 0, 94, 0, 101, 0, 131, 0, 129, 0, 110, 0, 122, 131, 99
and your example has no 2007_train.txt and data,can you provide it ?
can you anwser me ? thank you very much

When I make didn't have the AVX will have error

@ArtyZe Predictions are not drawing on the image after testing a newly trained model, is there any file that will store the predictions??

@ArtyZe Thanks for referning this repo , i have few queries

1. how different is ur repo with respect to any darknet
2. what was the weight file size before pruning and after pruning
3. Can i plot the weights values before pruning and after pruning to find out the difference

Can you share the pruned yolov2.weights?

你好，感谢您的分享。
請問可以說明一下關於L1 scales regularization 方法具體需要修改源碼那些地方嗎?