# TensorFlow CPU
pip install -r requirements.txt

Download pre-trained yolov4.weights file: https://drive.google.com/open?id=1cewMfusmPjYWbrnuJRuKhPMwRe_b9PaT

Download license plate trained custom weights: https://drive.google.com/file/d/1EUPtbtdF0bjRtNjGv436vDY28EN5DXDH/view?usp=sharing

Copy and paste custom.weights file into the 'data' folder. Copy and paste custom.names into the 'data/classes/' folder.

For custom model to work, change line 14 of core/config.py from coco.name to custom.names.

To implement YOLOv4 using TensorFlow, first we convert the .weights into the corresponding TensorFlow model files and then run the model.

# Convert darknet weights to tensorflow
## yolov4
python save_model.py --weights ./data/yolov4.weights --output ./checkpoints/yolov4-416 --input_size 416 --model yolov4 

# Run yolov4 on video
python detect_video.py --weights ./checkpoints/yolov4-416 --size 416 --model yolov4 --video ./data/video/input_video.mp4 --output ./detections/recognition.avi 

python save_model.py --weights ./data/custom.weights --output ./checkpoints/custom-416 --input_size 416 --model yolov4 

python detect_video.py --weights ./checkpoints/custom-416 --size 416 --model yolov4 --video ./data/video/input_video.mp4 --output ./detections/recognition.avi

To count number of objects per class use following command

python detect_video.py --weights ./checkpoints/yolov4-416 --size 416 --model yolov4 --video ./data/video/input_video.mp4 --output ./detections/recognition.avi --count

To count total number of objects, set the byClass=False in line 146 of detect_video.py and use the above command.

For this purpose we will use Firebase Realtime Database.

Step 1 - Create a Firebase account.

Step 2 - Create a Firebase Realtime Database.

Step 3 - Generate private key.

Note: For generating private key-

--> Click on "Service accounts".

--> Click on "Generate Private key" to get sdk file in json format.

Step 4 - Upload private key (json file) to the root directory of project.

Step 5 - Get the database url and paste it in ".env" file.

python detect_video.py --weights ./checkpoints/custom-416 --size 416 --model yolov4 --video ./data/video/input_video.mp4 --output ./detections/recognition.avi --crop

The crop flag will save detections to detections/crop folder and feed the cropped detections as input to Easy-OCR for character recognition purpose.

Note: The script is automated to do detection and character recognition simultaneously.