Real-Time Object Detection & Tracking System for ISR Drone Payload (Jetson Orin NX)

I’m looking for an experienced developer (or team) to help build a real-time Object Detection and Tracking system to be deployed on an ISR drone payload. The target platform is an NVIDIA Jetson Orin NX running Linux.

Core Requirements:

Real-time object detection (using YOLOv8 or equivalent)

Real-time object tracking (e.g. DeepSORT, ByteTrack, etc.)

Integration with live camera feed (USB/CSI camera)

Maintain object centered in frame (basic gimbal control or software-based digital panning)

Output real-time metadata (target position in frame, timestamp, confidence score, etc.)

Optimized to run efficiently on the Jetson Orin NX (TensorRT or similar acceleration)

Optional (Future):

Basic OCR capability (e.g. for license plates or signs)

Dual-camera architecture:

Wide-angle camera for detection

Narrow FOV or steerable camera for zoomed tracking

Distance/angle estimation using trigonometry or GPS data

Deliverables:

Complete source code (well-commented, modular, with setup instructions)

Real-time demo (could be with prerecorded video for testing)

Compatibility with Jetson Orin NX (Python preferred, C++ optional)

Brief documentation on architecture, dependencies, and how to retrain or tweak models

Ideal Skills:

Python / C++ for embedded systems

Deep learning with PyTorch, OpenCV, TensorRT

Jetson platform experience (Orin NX preferred)

Computer vision deployment and optimization

(Optional) Gimbal or servo control via PWM

Notes:

I will provide access to the hardware for remote testing if needed

Real-time performance and stability are key

Please include estimated timeline and cost in your proposal

Portfolio of similar projects is a plus