Hi there! This is an object detector. Given an image, it will localize the objects and tell which class the model thinks they belong to. There are only 80 classes that the model knows. Make sure to adjust your expectations :)
The detector is based on YOLOv3 and was trained on a large dataset of annotated images called COCO. It used to be the state-of-the-art among fast object detectors when I re-implemented it just for fun and glued to this project.
Under the hood, the webpage (front-end) will send your image to the cloud server (back-end). Then, the model makes detections and sends results for you to see here. If you are interested in details, check out my post on how I built it.
Please note that the results are saved on the server and inspected later on.