PictoPy Features

Gallery Application

Intelligent Photo Tagging: Automatically tags photos based on detected objects, faces, and facial recognition.
Traditional Gallery Management: Complete album organization and management tools.
Memories Feature: Automatically organize photos into meaningful collections based on location and date, with Google Photos-style presentation.

Object detection using YOLOv11 for identifying various items in images
Face detection and clustering powered by FaceNet.
People in videos (optional): faces found in video keyframes join the people recognized in your photos, so a person's page lists their videos too. Off by default; turn it on under Settings → Video Tagging. Videos tagged before it was turned on appear once they are scanned, from Scan videos there or the next folder sync.
Spatial Clustering: Groups photos by location using DBSCAN algorithm (5km radius)
Temporal Grouping: Organizes photos by date with monthly grouping
Reverse Geocoding: Identifies actual city names from GPS coordinates

Entirely offline: All data stays on your local machine.
No reliance on remote servers for processing.
Models are stored locally and customizable by the user.
Metadata export (optional, PNG only): saves tags, faces, embeddings, favourites and album names into your PNG files, so the work travels with the photos. Off by default; turn it on, or export the whole library once, under Settings → Image Metadata. See Metadata Export.

Lightweight SQLite database for storing image metadata, face embeddings, and album info.
Background image processing using asyncio for a smooth UI experience.

Search photos based on:
Detected objects
Recognized faces, one or several people at once, or a face from a photo or webcam (videos are included once people in videos is turned on and they have been scanned)
Embedded metadata
Semantic Search (optional, ~1.5 GB download): describe a photo in your own words — "beach sunset", "two people hugging" — and find it even if no detected tag matches the exact phrase. Powered by SigLIP2, running fully on-device via ONNX Runtime. Falls back automatically from tag search; no mode switch required.

Automatically creates meaningful photo collections inspired by Google Photos:

Location-based Memories: Groups photos taken at the same location (5km radius using DBSCAN clustering)
Displays as "Trip to [City Name], [Year]" (e.g., "Trip to Jaipur, 2025")
Uses reverse geocoding to show actual city names
Supports 30+ major cities worldwide
Date-based Memories: Groups photos by month for images without GPS data
Perfect for photos without location metadata
Organized chronologically