Exploring Vision with NVIDIA Jetson Orin Nano & Intel RealSense D457

I recently worked on a project where I combined the NVIDIA Jetson Orin Nano and the Intel Corporation RealSense D457 depth camera to develop a machine vision application. This setup allowed me to perform basic object detection, motion tracking, and utilize point cloud data, unlocking huge possibilities for real-world applications.

This project emphasizes the importance of depth perception and real-time analysis in machine vision. With the Jetson Orin Nano's incredible processing power and the depth camera's spatial data, I explored how these technologies can bridge the gap between perception and action.

Hardware

Intel RealSense D457 Depth Camera

• Resolution: Up to 1280x720 pixels
• Frame Rate: Up to 90 frames per second
• Depth Range: 0.2 to 20 meters
• Field of View: Wide field of view for enhanced depth perception
• Connectivity: USB 3.1 Gen 1 for fast data transfer

NVIDIA Jetson Orin Nano (8GB)

• GPU: 1024-core NVIDIA Ampere architecture GPU
• CPU: 6-core ARM Cortex-A78AE
• Memory: 8GB LPDDR5
• Performance: Up to 40 TOPS (Tera Operations Per Second)
• Connectivity: Gigabit Ethernet, multiple USB ports, and camera interfaces

Software

• NVIDIA SDK: Tools for accelerated AI computations and seamless deployment
• RealSense SDK: Software for processing depth data and integrating point clouds
• Ubuntu: Operating system used for development and deployment

Implemented Projects

• Object detection using a trained model
• Motion detection
• Point cloud viewing
• Depth image acquisition

Project Output