center_focus_strong AI Identification

How it works

We utilize state-of-the-art YOLO-based neural networks optimized for edge devices (Jetson Orin). The model is trained on a custom dataset of trash items found in urban environments, allowing for classification of multiple waste types in real-time.

The vision system runs continuously, scanning the environment and flagging potential targets for the navigation system to approach.

Technical Implementation

• Model Architecture: YOLOv8 (Ultralytics)
• Hardware: NVIDIA Jetson Orin Nano (CUDA Acceleration)
• Performance: < 3ms inference time per frame
• Datasets: Custom labeled dataset of urban trash