JRS_DetectCommand.txt

Look in the following file and make sure there is one class per line
C:\Users\jshea\Desktop\ML\MachineControlByObjectDetection\data\classes\obj.names

Start at the Anaconda prompt and enter the following:
conda activate yolov4-gpu

Change to the following directory:
cd C:\Users\jshea\Desktop\ML\MachineControlByObjectDetection

Put your weights file in the following directory:
C:\Users\jshea\Desktop\ML\MachineControlByObjectDetection\data

Rename your weights file to the following:
custom.weights

Convert YOLOv4 to TensorFlow before running detector:
python save_model.py --weights ./data/custom.weights --output ./checkpoints/custom-416 --input_size 416 --model yolov4


Run detector against sample video input.
All defaults are used.
Relay will be used if inserted into USB port.
python JRS_DetectorSorter.py

Run detector against webcam input. 
In this case selecting webcam input from webcam at index 0.
Score is set to 1/2 sensitivity. (.1 is most sensitive.) (.9 least sensitive)
Count flag sends relay status info to the console along with info about objects detected.
Output flag sends a video of detected objects to disk.
Relay will be used if inserted into USB port.
python JRS_DetectorSorter.py --video 0 --score .5 --count --output ./detections/results.avi


See all the options with the following command.
python JRS_DetectorSorter.py --help