-
I am using yolov4-p5 with an inference size of 640. I get slightly better results if I use --no-autoanchor
I used to use autoanchoring and copy the generated anchors from the terminal for use with…
-
A quick question for experts in this group.
Q: How YOLO generates the predefined Anchor boxes for a new data set? We do not define anything for Anchor boxes and the framework generates itself. How…
-
```
./darknet detector train data_file cfg_file -gpus 0,1,2,3 -map
CUDA-version: 11020 (11020), cuDNN: 8.1.0, CUDNN_HALF=1, GPU count: 4
CUDNN_HALF=1
OpenCV version: 4.5.1
0 : compute…
-
/usr/local/cuda/bin/nvcc -Iexternal -I/opt/conda/lib/python3.6/site-packages/torch/include -I/opt/conda/lib/python3.6/site-packages/torch/include/torch/csrc/api/include -I/opt/conda/lib/python3.6/site…
-
### my dataset img
classes:10
train : 301
valid : 144
all img size is 96x96
### calc anchors
1. 608x608
 video streams is compared with the processing of a single video stream. fps=single fps/n, why? GPU utilization is less than 30%, cpu utilization is less tha…
-
hi
i use u5 to train my own dataset (with pretrain weigts: --weights weights/yolov4l-mish.pt). I notice the losses are very low as shows like
` Epoch gpu_mem GIoU obj cls t…
-
I use the latest darknet commit for training cspx-p7-mish_hp.cfg, I change the classes and filters before each yolo layer
filters=(5+classes)*3, and start the train but the error log
```
filters=…
-
I train my own model with the default value you provide. It show me that: