Object Detector (2022)
The Object Detector is a prototype for a smart stock tracking system. The prototype comprises an ESP32-CAM, an FTDI module and a micro-USB cable. A protective casing, consisting of a plastic box and an MDF plate to secure the microcontroller, features openings for the micro-USB cable and camera connections.
To control the module, a C++ program was written in the Arduino IDE. While models can be trained to recognize objects, this process is complex and time-consuming for a short project. Therefore, this project utilizes COCO-SSD, a machine learning model for object recognition in images. The COCO dataset comprises over 330,000 images with over 1 million objects classified into 80 categories, including people, vehicles, and buildings.
A demo of the system can be viewed below.