Tutorials Yolo_Object_Detection for custom data:
Method 1: Use LabelImg -> export file txt
Method 2: Use https://www.makesense.ai/
Note: id classes as number object.
!git clone 'https://github.com/AlexeyAB/darknet.git' 'PATH_to_darknet'
!git clone 'https://github.com/AlexeyAB/darknet.git' '/content/drive/MyDrive/yolov3_v1/darknet'
!make
creating-files-data-and-name.py
creating-train-and-test-txt-files.py
Then Upload file darknet53.conv.74 inside custom_weight
!python3 /content/drive/MyDrive/yolov3_v1/images/creating-files-data-and-name.py
!python3 /content/drive/MyDrive/yolov3_v1/images/creating-train-and-test-txt-files.py
GPU=1
CUDNN=1
OPENCV=1
Then save, upload darknet
Line 6, 7: delete #
Line 20: Max_batch = number class * 2000
Line 22: Step = 90%, 110% max_batch
Find [yolo]: edit: classes, filters
Save name: yolov3_custom.cfg then upload cfg.
!darknet/darknet detector train images/labelled_data.data darknet/cfg/yolov3_custom.cfg custom_weight/darknet53.conv.74 -dont_show