ttdd11
ttdd11
Are the scaled models (ie. [yolov4-p5.cfg](https://raw.githubusercontent.com/AlexeyAB/darknet/master/cfg/yolov4-p5.cfg)) supported? Also, in the .cfg for the yolov4-csp has a "logistics" activation, is it okay if that is changed to linear in the config...
Amazing repo and very interesting work. I have some questions about the data formatting to introduce 3D data. This may sound very simple so I apologize in advance. When building...
This repo works great! I think a nice addition would be to give some context for the scores. Demo.py shows the top ten results, but sometimes this isn't helpful. Really...
I'm looking to checkout the 2D work here without the IK and 3D projection. How challenging do you think it would be to get this going on Windows just in...
Typically when using TensorRT I see significant speed improvements but on my card I haven't really seen any (2080 ti). The bounding boxes look good in all versions, so I...
I am having problems downloading the 418 model. Is it possible to add it to google drive?
Amazing contribution to the community. Do you believe this method will perform similarly well for fine grained keypoints such as coco-whole body (which has hands and facial keypoints), or would...
Great work. I'm porting this to onnx import and it's working okay. Quick question on the ouput, why use "yolo-det";//layer82-conv,layer94-conv,layer106-conv Instead of the commented code at the end. I have...
How do you think a batch size of 8 versus the batch size you used affect the final output? Have you experimented with batch size at all to see the...
What was the rational for not flipping images (vertically) in the data augmentation? I think even better results could be achieved with that.