{"id":67997,"date":"2022-07-14T21:06:03","date_gmt":"2022-07-14T13:06:03","guid":{"rendered":"https:\/\/www.seeedstudio.com\/blog\/?p=67997"},"modified":"2023-05-18T17:45:15","modified_gmt":"2023-05-18T09:45:15","slug":"build-an-nvr-camera-system-with-frigate-monitor-your-security-cameras-with-locally-processed-ai","status":"publish","type":"post","link":"https:\/\/www.seeedstudio.com\/blog\/2022\/07\/14\/build-an-nvr-camera-system-with-frigate-monitor-your-security-cameras-with-locally-processed-ai\/","title":{"rendered":"Build an NVR camera system with Frigate! Monitor your security cameras with locally processed AI"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n

Today we wanna share a super cool open-source project to help you build up an NVR system with AI detection, and everything will be processed locally! <\/p>\n\n\n\n

Frigate<\/a> is a complete open source and local NVR designed for Home Assistant with AI-powered object detection(Kudos sents!?). It uses OpenCV and Tensorflow to perform real-time object detection locally for IP cameras. Frigate brings a rich set of features including video recording, re-streaming, motion detection, and supports multiprocessing.<\/strong> Leveraging Google Coral TPU USB accelerator<\/a> for 100+ FPS object detection, you can run advanced AI analysis with an Odyssey Blue, an Intel Celeron J4125 powered powerful Linux mini PC<\/a>, to determine if the motion is actually a person, car, or any object you want to detect. With Frigate\u2019s local processing, there is no need cloud fee anymore for storing your personal camera footage and the analysis record. <\/p>\n\n\n\n

\"\"\/
?Frigate<\/figcaption><\/figure>\n\n\n\n

Why Frigate<\/a>?<\/h2>\n\n\n\n

Traditional NVRs have higher false positive rates and can require hours of fine-tuning to reduce false positive rates because they rely on simple motion detection. Leveraging Google Coral TPU<\/a>(USB accelerator and M.2 Module) for object detection, you can run advanced AI analysis with a mini PC to determine if the motion is actually a person, car, or any object you want to detect. With Frigate’s local processing, there is no need cloud fee any more for storing your personal camera footage and the analysis record.<\/p>\n\n\n\n

All processing locally <\/strong><\/p>\n\n\n\n

No need to pay for sending your personal camera footage to the cloud for analysis.<\/p>\n\n\n\n

Never miss a frame of detection <\/strong><\/p>\n\n\n\n

Frigate can run over 100 object detections per second, so it never misses a frame. Frigate tracks objects in real-time and can determine the exact time a person starts walking up your front steps or a car enters your driveway.<\/p>\n\n\n\n

Design for Home Assistant, seamless integrate into your smart home system<\/strong><\/p>\n\n\n\n

Make your home eyesight by integrating object detection into Home Assistant, OpenHab, NodeRed, or anything with MQTT support. Frigate integrates directly into Home Assistant’s media browser, provides a low-latency camera entity, and exposes real-time sensors and switches to power automation and notifications to your heart’s content.<\/p>\n\n\n\n

\"\"
?Frigate: Integrate with Home Assistant and other automation platforms<\/figcaption><\/figure><\/div>\n\n\n\n

Easily check the target camera<\/strong><\/p>\n\n\n\n

Dynamically renders cameras with active detection by Birdseye view. <\/p>\n\n\n\n

\"\"\/
?Frigate<\/figcaption><\/figure>\n\n\n\n

What hardware do you need for Frigate? <\/h2>\n\n\n\n