search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.65k stars 7.96k forks source link

Custom object detection training doesn't start. No errors shown. #8351

Open sreddy-es opened 2 years ago

sreddy-es commented 2 years ago

I have been trying to run custom object detection and followed the exact steps to get the data in the right format and to download the weights. The training never starts and the script ends without any warnings and errors please advice as to what I can do here.

training command:

./darknet.py detector train data/obj.data cfg/yolov4-tiny-obj.cfg yolov4-tiny.conv.29 -dont_show -map -mjpeg_port 8000
CUDA-version: 11000 (11000), cuDNN: 8.0.5, GPU count: 1
 OpenCV version: 3.2.0
 0 : compute_capability = 370, cudnn_half = 0, GPU: Tesla K80
net.optimized_memory = 0
mini_batch = 1, batch = 1, time_steps = 1, train = 0
sreddy-es commented 2 years ago

Had to use the exact compute capability of a K80 which is 3.7 in the makefile and had to recompile recompile.

Modified line 20 in Makefile: ARCH= -gencode arch=compute_37,code=sm_37

prudhvi-charan commented 2 years ago

Did you check the train.txt file? The training won't happen if it's empty.