Today we are going to build our own real-time AI surveillance camera using a raspberry Pi and the OpenCV AI kit. We will adapt this system to detect any object, like packages on your doorstop, with Roboflow. This Raspberry Pi CCTV camera will be able detect a person using a MobileNet SSD at an astonishing 30 FPS as well as sound an Alarm to stop any impending theft.
⭐ Train models with Roboflow: https://roboflow.com/as1
⭐ Free computer vision datasets (like packages): https://roboflow.com/as-datasets1
⭐ GitHub + OAK Course: https://augmentedstartups.mykajabi.co...
Links Mentioned in the Tutorial:
Tools used in the Tutorial:
► Raspberry Pi 4 - https://amzn.to/3fhSI7c
► Buy OAK Device - http://bit.ly/GetOAKNow
► Speaker - https://amzn.to/3fqnoDn
► 3.5mm Aux Cable - https://amzn.to/2ITU2B8
► LED - https://amzn.to/3lLzUj1
► Samsung Note Ultra - https://amzn.to/38Wjj8t
► Roboflow - https://roboflow.com/as1
The Problem that we are trying to solve is, that there are not much real time raspberry pi CCTV camera solutions online and if they are real-time, then the processing is not an embedded solution, its done on a server, PC or cloud.
0:00 - Introduction
1:03 - Camera Build Overview
2:06 - Requirements
2:56 - Connecting the RPI to LED and Speakers
3:23 - 3D Printing Part
5:20 - Download the Repo
6:40 - Coding Tutorial and Explaination
18:07 - Integration - Putting it all together and Mounting
18:46 - Testing
18:57 - Training
20:25 - Roboflow Data Augmentation Walkthrough
Another option is to use something like a Neural Compute Stick or Google Coral, which is great but their form factors are slightly bulky especially when we are trying to minimize our security camera size.
Instead we are going to try using the OpenCV AI Kit attached to almost any raspberry pi even a raspberry pi zero for a small footprint surveillance camera. So let's get straight into it!
Join our mailing list to receive the latest news and updates from our team. You'r information will not be shared.