Successful use of FOMO model for Vehicle Detection

Just wanted to share a project I recently completed, which utilized the OpenMV H7 Plus and a trained Edge Impulse model to detect vehicles.

Project Links:

The device is an IoT vehicle counter, which can be deployed to monitor any road to count the number of vehicles that drive past it for traffic monitoring purposes.

The OpenMV was used to collect a training dataset of hundreds of images that include vehicles, and Edge Impulse was used to train a FOMO object detection model with this dataset. The model was able to reach a 95% accuracy on the testing split of the dataset.

The trained model was exported and uploaded onto the OpenMV Cam, and utilized in a Python script to process newly captured images in real time. The processing happens quickly enough to allow for the detection of each vehicle as it drives past the device.

This is my second project completed using the OpenMV Cam and Edge Impulse, and I am impressed by how well they operate and perform together!


Nice work! @davidtischler_edgeim check this out!

I saw this project when you posted it the other day @rhammell, but had not read it until just now. WOW! This is extremely impressive and very well documented, great job!!