SyGoing
您好，大佬，请问您有没有统计过如题的数据？，精度上和原始caffe-mtcnn有没有更强，速度及显存占用有没有提升？
I have use the same net structure but when training it to 10000iter ,the loss is always 9.***，is there some tricks of training the mobileFaceNet?
yulibcafe,cpp ,when you use the caffe model, you first revert it and the input shape change from caffe(nchw) to tf mode(nhwc), why you do that instead of using it as...
### Search before asking - [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussions) and found no similar questions. ### Question when I train my own data with YOLO8, I...
我用大佬的NCNN工程nanotrack测试自己的一个视频，发现跟踪框不是很稳定，抖动比较厉害。同样测试了网上的一个ncnn版本的LightTrack感觉效果要好很多，请问如何提升精度，我看数据集貌似只用GOT，LightTrack使用了多个数据集混合训练，网络结构也有比较大差异
cuda out of memory: the log is like this 2019-05-09 15:24:17.105701: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_driver.cc:936] failed to allocate 6.25G (6712326144 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY 2019-05-09 15:24:17.534339: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_driver.cc:936] failed to allocate 5.63G...
我把他转到windows下 发现parse onnx文件时出错，fd返回3
The training data can be marked by labelme? , and the training png is index picture? And, If I modified your net to make the output's tensor's shape equal input,...
Long times ago,I also want to make a caffe which contains all kinds of implementation. You do a good job, I 'm very interested to your job and if possible...