Computer Vision Architecture Review – RGB-D 3D Reconstruction for Livestock

We are building an agritech platform that uses camera data to estimate physical traits of livestock, including body condition and structural measurements.

We are looking for a computer vision expert to review the architecture of the camera-vision part of the system. This is a fixed-price architecture review, not a development or implementation project.

The system uses handheld RGB cameras as well as RGB-D stereo depth cameras, specifically Intel RealSense D400 series, for higher-quality fixed-position capture.

You will receive a written technical specification covering the sensor setup, capture strategy and data model. Your job will be to review the proposed approach and give us an independent technical opinion on what works, what may cause problems, and what you would change.

The main focus is dense 3D reconstruction and structural measurement from depth and surface geometry. This is not an object detection or image classification project.

Experience with photogrammetry, stereo vision, RGB-D systems, 3D reconstruction or point-cloud based measurement is therefore much more relevant than general computer vision experience.

The final output will be a written report with recommendations against a defined set of review criteria.

Two rounds of follow-up clarification are included after the initial report.