GuestGallery

Overview

GuestGallery is a Python face-recognition tool built to solve a very concrete problem: allocating 2000+ wedding photos across guests automatically, so each person gets exactly their own photos without anyone manually sorting a shared album.

Under the Hood

At 2000+ images, brute-force face comparison doesn't scale — the pipeline needed to batch-process efficiently, handle varied lighting and angles from a real wedding shoot, and still produce accurate per-guest matches rather than false positives that annoy people expecting their own photos.

What I Built

• Face recognition pipeline detecting and matching faces across the full 2000+ image set
• Guest enrollment flow — register each guest with one or more reference photos
• Bulk automated allocation — every photo assigned to the guests who appear in it
• Personal galleries — each guest receives their own curated photo subset
• Efficient batch processing tuned to handle large image sets without prohibitive runtime
• Structured output — per-guest folders or shareable links

Tech Stack

• Language: Python
• Face recognition: face_recognition, dlib
• Image processing: OpenCV, PIL
• Storage: Local or cloud output storage

Key Decisions

• Batch-processed the image set rather than comparing faces one-at-a-time in real time, since 2000+ images made a naive brute-force approach impractical at runtime
• Required guest enrollment with reference photos before allocation rather than attempting unsupervised clustering, trading a small upfront step for materially more accurate per-guest matches
• Tuned the matching threshold to favor precision over recall, since a guest missing a photo they appeared in was a far smaller problem than guests receiving photos of strangers

Outcome / Impact

Eliminated fully manual sorting of thousands of wedding photos — guests received their personal galleries in minutes instead of the hours it would have taken by hand.