AI Bird Identifier PART 3!! Identify birds on YOUR bird feeder using a RASPBERRY PI and AI

Welcome to part 3 of my project to build an AI-driven, Raspberry PI-based, bird identifier that can be used to identify bird species on my bird feeder!

In part 2, I built a custom AI data model, trained from images captured from an IP webcam facing my bird feeder.

This time, I apply that custom data model to a live video feed from my IP webcam so that I can identify birds in real time!

Links to things you see in the video (shop around!! These are just what I used!!):
Raspberry PI 5 - https://shop.pimoroni.com/products/ra...
Webcam (This is the one I used but any weatherproof camera that captures full HD would be fine) - https://www.amazon.co.uk/gp/product/B...
My code shown in this video: https://github.com/jeffspiinthesky/UK...

Other videos referenced in the video:
Bird classification Pt2 -    • AI-driven BIRD IDENTIFICATION on RASP...  
Take control of VIDEO streams from YOUR IP camera -    • Take control of the VIDEO streams fro...  

Chapters:
00:00 - Intro
00:34 - Titles
00:41 - Like / Subscribe / Notify / Comment
01:14 - Why I didnt use Frigate NVR
03:08 - Intro to my solution
03:31 - Clone code from GitHub
04:02 - Code walkthrough (webcam.py)
06:18 - Code walkthrough (ffmpeg.py)
10:56 - Code walkthrough (ai_analyser.py)
12:50 - Code walkthrough (visualiser.py)
14:01 - Code walkthrough (exceptions.py)
14:15 - Code walkthrough (app.py)
17:04 - Code walkthrough (Dockerfile)
19:36 - Code walkthrough (docker-entrypoint.sh)
20:17 - Building the container
21:13 - Docker compose config file
23:14 - Start container from command line
24:10 - Stop container from command line
24:28 - Run container from Open Media Vault
25:25 - Demo - Opening RTMP stream in VLC
26:10 - Code walkthrough (index.html)
26:32 - Add index.html to NGINX container config
26:46 - Demo - Opening HLS stream in my browser
27:22 - Conclusions and future enhancements
29:07 - Outro
29:40 - Closing titles

Equipment used:
Canon M100 camera with 13-55mm kit lens
Shure MV7 mic
Elgato collapsable green screen
Neewer GL1 Pro keylight
Nanlite Pavotube 30C lights
Original video capture using OBS
All video editing using Kdenlive
All image editing using GIMP