Simple Object Recognition with the UNIHIKER K10

This object recognition system operates entirely on the UNIHIKER K10 development board, leveraging its integrated ESP32-S3 microcontroller, built-in camera, color display, speaker, microphone, and onboard PSRAM. Instead of relying on heavy cloud servers or complex, resource-intensive neural network frameworks, the system processes visual input locally on the edge. The built-in camera captures real-time video frames, which are temporarily stored and managed using the board's PSRAM to handle the memory demands of image processing. Rather than running deep learning models, the firmware analyzes visual signatures and region brightness directly from the captured pixel data to detect and identify objects. When a targeted object or visual signature is recognized based on these brightness and pattern parameters, the UNIHIKER K10 can trigger immediate feedback locally. This feedback is delivered through its integrated color display—which can render visual indicators or status data—and its onboard speaker or microphone for audio alerts or cues. Coded using C++ through the Arduino IDE, the system combines hardware-level camera management with lightweight computer vision heuristics, demonstrating how powerful edge-based classification can be achieved on a compact microcontroller architecture.