Ming Qi
Ming Qi
I've tried to train model with DCNv2 via apex mix-precision method, but occurred a dtype bug: ``` "deform_conv_forward_cuda" not implemented for 'Half' ``` Does it mean that I cannot train...
As is described above , I wonder the speed FCN can achieve , but the paper doesn't show that......
When I try to apply apex for mix-precision training , error appears as follow" ``` expected scalar type Float but found Half (data at /home/bit530/anaconda3/envs/torch1.1/lib/python3.7/site-packages/torch/include/ATen/core/TensorMethods.h:1821) ``` thus I convert all...
Is it possible to implement corner pooling wtih pure python? Or cuda/cpp extension is necessary ? (I am not good at cpp/cuda programming ... and consider sorting to python implemention...)
Could you provide me with a sample file for evaluation (such as sample.zip for submission), is it in dota form or coco form? The test email has not responded to...
Have anyone tested on ICDAR15 dataset? How about the effect(Hmean or F1 score)?
there should be limitation on variable `weights `, how to set it? Or which range is suitable? I apply GHMC loss directly on rotated-yolo for conf loss calc, loss explosion...
之前想等到IC15数据集刷到比较高再开源，目前卡住了估计一时间上不去放弃ryolo的模型了，感兴趣的可以直接用我的[code](https://github.com/ming71/rotate-yolov3)