Segregation of Dehusked Arecanut using Artificial Intelligence Technique on Raspberry Pi

Real-time object identification is of significantly importance across many industries, including agriculture. Since it has the potential to optimize efficiency and accuracy in tasks such as grading and quality control. This research introduces a novel methodology for real-time detection of arecanut objects utilizing the YOLOv5 model on a Raspberry Pi platform. A thorough dataset of arecanut was collected and compiled, wherein the arecanut were categorized into several categories based on their husk characteristics. The YOLOv5 model underwent training using the provided dataset, and data preprocessing was conducted to optimize precision accuracy to its highest potential. Subsequently, the model was implemented on the Raspberry Pi, considering the limitations imposed by the hardware. The model was optimized for the processing resources of the Raspberry Pi by the implementation of quantization techniques, resulting in improved efficiency for real-time inference. By conducting thorough experimentation and evaluation, the study provides evidence of the efficacy of proposed methodology, which enables real-time object detection to grade arecanut on the Raspberry Pi platform. The proposed solution exhibits the potential to bring about a revolutionary transformation in arecanut processing and comprehensively enhance agricultural practices. This is achieved via the provision of a tool that is both cost-effective and efficient for the purpose of object-detecting duties.