Waste management is a growing concern in today’s time and efficiently classifying is a very crucial aspect in increasing rates, reducing pollution, and ensuring stability. The prototype is a lightweight and low-cost model that segregates waste using the YOLOv5n model for real-time object detection locally. We combine the model with Arduino Uno to make a low computational overhead waste-sorting device that categorizes waste such as “Paper,” “Plastic,” and “Organic.” Its application comprises small-scale areas such as campuses and offices. The model trained over 50 epochs reports precision and recall at 0.795 and 0.739, respectively. Class-specific precision was the highest for “Paper” at 0.901 and “Plastic” at 0.857. This prototype promotes automation and sustainability toward efficient waste management practices that can be done locally without relying on cloud support or Internet connectivity.