Computer Vision Engineer – Retail Shelf Image Registration & Multi-Image Super-Resolution

**Project Overview** We are building an automated workflow that creates a single high-resolution image of a retail shelf (bay) from multiple overlapping smartphone photos. **Objective** Given: * One overview image of an entire retail bay (optional) * 20–40 overlapping close-up images captured from approximately the same distance Develop software that: * Automatically detects where each image belongs. * Aligns all images with sub-pixel accuracy. * Corrects perspective distortion. * Blends overlapping regions seamlessly. * Produces a single high-resolution composite image suitable for planogram review and retail analytics. * Supports batch processing of multiple bays. **Preferred Technology** * Python * OpenCV * SIFT / ORB / AKAZE feature matching * Homography estimation * Image registration * Multi-band blending * Camera calibration * Photogrammetry experience is highly desirable. **Bonus Skills** * COLMAP * OpenMVG / OpenMVS * CUDA * PyTorch * GPU acceleration * Experience with large image mosaics **Deliverables** * Well-documented Python code. * Command-line tool for batch processing. * Example outputs using our sample image sets. * Documentation describing recommended image capture procedures. Please include examples of previous projects involving panorama stitching, photogrammetry, image mosaics, document stitching, aerial imagery, or other computer vision work involving image registration.