Data is Annotated using Labelme and is available in yolo format with txt files
      ├── data
      │    ├── obj (Train dataset)
      │    │   ├── Knife
      │    │   ├── Handgun
      │    │   └── Shotgun
      │    ├── test (Test dataset)
      │    │   ├── Knife
      │    │   ├── Handgun
      │    │   └── Shotgun
      │    ├── train.txt  (Train label)
      │    └── test.txt   (Test label)

  Training folder contain important files
      ├── training
      │    ├── obj.data
      │    ├── obj.names
      │    ├── yolov4-tiny.config
      │    └── yolov4-tiny.conv.29  

 # train your custom detector! (uncomment %%capture below if you run into memory issues or your Colab is crashing)
 # %%capture
 └── !./darknet detector train training/obj.data training/yolov4-tiny.cfg training/yolov4-tiny.conv.29  -dont_show -map
 
 # run your custom detector with this command (upload an image to your google drive to test, thresh flag sets accuracy that detection must be in order to show it)
 ├── !./darknet detector test training/obj.data training/yolov4-tiny.cfg /mydrive/yolo/yolov4/darknet/backup/yolov4-tiny_4000.weights          /mydrive/yolo/yolov4/darknet/data/test/Knife/7bc500661f0ea2c7.jpg -thresh 0.5
 └── imShow('predictions.jpg')

   ├── tflite.ipynb