ZJ
`gary_img = cv2.cvtColor(roi, cv2.COLOR_RGB2GRAY)` why not `gary_img = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)`
我有个疑问，为了检测远距离的小人脸，主页建议采用大的训练尺寸，例如640x480，我理解的是随机crop之后再resize，如果resize尺寸小一点，即输入尺寸小一点，这样数据集中训练的人脸不就变小了，比较符合远场景的小人脸，为什么建议采用大的训练尺寸呢，大了之后人脸其实缩小的比例不够，太小的人脸就检测不到了
I use the create_dataset.py to make lmdb dataset. But get this error. When I run this script in my own computer, it works.Why? The error like [this](https://github.com/jnwatson/py-lmdb/issues/100)
有人crnn每个batch 宽度不定长 训练过程中期loss nan ，崩了吗？ @wenlihaoyu 请问大佬遇到过了么
[https://github.com/zeusees/HyperFT/blob/master/Prj-Linux/CMakeLists.txt#L29](https://github.com/zeusees/HyperFT/blob/master/Prj-Linux/CMakeLists.txt#L29) linux下视频输入换成本地视频，结果保存在一个avi中，去掉 /usr/lib/x86_64-linux-gnu/libsqlite3.so编译可以成功，但是一运行就崩
Great work!!! Can you give some examples for multi-task. I think the examples in 'experiments' are all single tasks. Is there anything wrong with my understanding?
[https://github.com/xiangdeyizhang/FaceTrack_ncnn_HyperFT/blob/master/CMakeLists.txt#L29](https://github.com/xiangdeyizhang/FaceTrack_ncnn_HyperFT/blob/master/CMakeLists.txt#L29)
测试时候贴的性能对比，都是320x240,640x480的