• Python 3.6+
• Matterport Mask R-CNN
• OpenCV

• Copy the file pet_recognition.py to the directory with the Mask R-CNN
• Create a test_images directory
• Add to the directory test_images video file on which you want to recognize the cat
• Change the video source file path in the VIDEO_SOURCE in the pet_recognition.py
• run the script through a terminal or console command

Linux

python3 pet_recognition.py

Windows

python pet_recognition.py

• the video stream starts and breaks into frames
• the neural network searches for a pet at each frame (the first appearance of the pet on the video and its subsequent movements are taken into account)
• Count the number of frames on which there is no pet.
• notification of the appearance in the frame of a pet
• notification of pet leaving the frame