Computer-Vision Photo Enhancement Pipeline for a High-Volume Listing Photographer

Overview

Built an automated image-processing pipeline for a real estate photography studio shooting 40+ properties a week, where manual editing had become the bottleneck between shoot day and listing day. The system uses computer vision to correct exposure and perspective distortion, remove minor clutter, and enhance sky and lighting consistently across an entire shoot in one batch. Photographers upload raw images through a simple web interface and receive listing-ready photos back within minutes instead of the prior next-day turnaround from a manual editing service. The pipeline flags any images needing human review rather than attempting a fully automated finish on every shot. Manual photo editing had become the bottleneck between shoot day and listing day for a studio shooting 40+ properties a week, with a next-day turnaround from a manual editing service slowing down every listing. We built an automated computer-vision pipeline that corrects exposure and perspective, removes minor clutter, and evens out sky and lighting across an entire shoot in one batch, flagging only the images that genuinely need a human look. We started by collecting a large sample of the studio's past raw-to-edited photo pairs to understand their exact editing style, then built and tuned the pipeline against that style rather than a generic preset. After a pilot run alongside the studio's existing manual process, we compared outputs side-by-side before cutting over fully. Photo turnaround dropped from a next-day service to under 30 minutes, letting listings go live the same day they're shot instead of waiting on external editing capacity.