How to Use Your Raspberry Pi to Detect Objects (TensorFlow + Camera Module)

I will demonstrate how we can use our Raspberry Pi Model B+ to detect ANY object captured in the camera frames or video. Specifically we create a model that looks for instances of guitars within pictures and returns a probability that the image contains a guitar. We use TensorFlow, Docker, and the Macgyver API to string together these technologies and deploy them to our edge device. This video will walk you through each step so that you will be able to do the same.

Specifically this demo uses a generic camera module attached to a Raspbery Pi 3 Model B+. We write a node.js script which uses the Raspi Cam programming interface to take a picture every 3 seconds. The program uses Google Cloud Utils to upload that image to a Google Cloud Storage bucket. We then make a request to the macgyver api to run the machine learning model and generate a prediction on the given input image. The program then responds by playing an audio file if either the guitar is detected or there is no guitar detected.

Node.js Script
https://github.com/tmoody/raspberry-pi-object-detection/blob/master/guitar.js

#tensorflow #raspberrypi #computervision

Camera Module Used
https://www.amazon.com/gp/product/B07DNSSDGG/ref=oh_aui_detailpage_o07_s01?ie=UTF8&psc=1

Wayixia Batch Download Chrome Extension
https://chrome.google.com/webstore/detail/batch-image-downloadfull/ahajhopfbfpekcljjjppolcmapaidldc?hl=en

Macgyver Tensor Flow Docker Container
https://hub.docker.com/r/macgyvertechnology/tensorflow

Music Credit: https://www.bensound.com/