TECHNIQUE FOR USING COMPUTER VISION TO ANALYZE ROOT CROPS IN THE BACK OF A TRUCK

Sorting sugar beet roots by size is essential for efficient storage and subsequent industrial processing. Standard methods for sorting vegetables by size in storage facilities with large volumes of root crops are not applicable to sugar beet roots. A method for using computer vision for contactless size and contamination determination of sugar beet roots in the back of a truck at a sugar mill's beet processing station is presented in this article. A prototype system has been developed, including a hardware and software system (microcontroller, laser rangefinder, RGB camera, additional lighting sources, LED display), a server, and an automated operator workstation. The system is designed to automate the process of assessing the size characteristics of root crops for their distribution into piles for short-term, medium-term, or long-term storage, as well as for processing. A computer vision system operating algorithm, which includes four sequential stages, is proposed. During the "Search" stage, a vehicle is detected under the camera using a laser rangefinder. Multi-level detection of root vegetables in the image is performed using a neural network implemented in Python. Calculation optimization is performed during the "Recognition" stage. Root vegetable dimensional parameters are determined based on camera data using mathematical formulas during the "Analysis" stage. A structured report based on the original and processed root vegetable images, along with statistics on their quantity and size classes, as well as root vegetable contamination, is generated during the "Recording" stage.