Computer Vision Expert for Custom YOLO Model Training & Target Tracking

Job Title: Computer Vision Expert for Custom Target Tracking Optimization ​Job Description: I am looking for an expert in Computer Vision and Deep Learning to develop and train a high-precision object detection model (YOLO-based) specifically for Uncharted 4 Multiplayer. ​The goal is to create a professional-grade, stable tracking system with the following requirements: ​Adaptive Bounding Box: The system must maintain a stable tracking box restricted to the target’s upper body (from waist to head). This box must be adaptive; it must precisely adjust and scale according to the target's distance (e.g., focusing specifically on the head area for long-range targets). ​Target Exclusivity: The model must identify active human players only. It must be strictly trained to ignore: ​Map elements, static assets, and environmental clutter. ​Visual distractors like objective markers (e.g., the red 'King of the Hill' capture zone). ​Non-player objects that resemble human silhouettes, such as statues, posters, and paintings. ​High-Performance Training: I require a robust training process using: ​Data Augmentation: Incorporating various lighting conditions, motion blur, and occlusions (partially hidden targets). ​Negative Samples: Extensive training on "friendly" markers and background elements to ensure zero false-positive detections. ​Movement-Based Validation: The model should prioritize active human movement patterns over static shapes to avoid false triggers. ​Integration Readiness & Documentation: ​The final model must be stable, modular, and ready for integration into a Python-based control system. ​I require the code to be clean, well-commented, and accompanied by brief documentation on how to fine-tune the model for future adjustments. ​Latency Optimization: The model must be optimized for high frame rates (FPS) and minimum latency to ensure real-time responsiveness. ​Application Requirements: Please highlight your experience with custom YOLO training, object tracking stability, and your approach to handling environmental noise in high-motion environments. If available, please provide a video demo or examples of your previous work in target tracking stability