LC044/TrailSnap

TrailSnap – AI‑powered travel photo album

What it is – TrailSnap (行影集) is a self‑hosted web application that turns a personal photo collection into an intelligent travel journal. It combines classic photo‑management features (albums, tags, map view) with AI‑driven analysis such as OCR on tickets, face and scene recognition, and large‑model chat that can generate travel diaries or short VLOG scripts.

Key capabilities

Feature	What it does
Smart album	Automatic classification by scene, person, location; OCR on tickets; searchable metadata.
Travel itinerary extraction	Detects train tickets, itineraries and other travel documents in images and pulls out structured data.
AI Agent	One‑line prompt to a large language model that writes a travel diary; hooks for Claude, OpenClaw, etc.
CLI & Skills	Command‑line interface and plug‑in system for automated tasks (e.g., calling external AI services).
Yearly report	Generates a visual summary of a year’s trips – photo wall, city map, mileage, timeline.
“On this day”	Shows photos taken on the same calendar date in previous years, sorted by AI‑assigned relevance score.
Map & timeline view	Interactive map with pins, trajectory view, and smooth scrolling timeline.
External folder sync	Scans a user‑specified folder for new photos/videos and indexes them automatically.
Future‑planned AI	Video‑clip generation, AI‑enhanced photo retouching (not yet released).

How to run it

• Desktop installer – pre‑built binaries for Windows/macOS/Linux are provided on the Releases page.
• Docker – a one‑line script (curl … | bash) installs Docker images for the server, AI micro‑service, PostgreSQL (with pgvector for vector search), and the frontend. A full docker‑compose.yml is supplied for manual deployment, with environment variables for DB credentials, AI service URL, and public URL.
• Source – instructions for building from source are linked in the docs.

Tech stack (inferred from README)

• Backend: Python (AI micro‑service) + PostgreSQL with pgvector for embedding‑based search.
• Frontend: Likely a modern JavaScript framework served behind Nginx (the frontend container).
• AI integration: Calls out to an LLM endpoint (AI_API_URL) and uses OCR/vision models for ticket and face detection.

Who it’s for

• Families or individuals who want a private, self‑hosted photo hub that automatically extracts travel information and can produce narrative summaries.
• Users comfortable running Docker or installing a desktop client; the project emphasizes data staying “truly yours”.

Community & support

• Official website (https://trailsnap.cn) with docs, demo site, and a Chinese‑language QQ/WeChat community.
• The repo is open‑source; contributions are welcomed via pull requests.

Bottom line – TrailSnap is a genuine AI‑enhanced photo‑album application focused on travel memories, offering both ready‑to‑use containers and a desktop installer, and it leverages large‑model APIs for diary generation and image analysis.