WildTrace — Every Small Habitat Has a Story
This is a submission for the Hacktoberfest Open-Source AI Challenge — Week 1: Touch Grass. What I Built I spend a lot of time in front of a computer building things with AI. Between experimenting with models, developing applications, and trying to turn ideas into working products, it's easy to spend hours looking at a screen without paying much attention to what's happening outside. When I came…
This project, WildTrace, is an AI-powered micro-habitat observation platform that encourages users to engage with nature by documenting small natural environments and comparing photographs of the same place over time. The platform's interface is nature-inspired, integrating habitat observations, comparisons, and field missions. WildTrace aims to motivate users to go outside, observe a small part of nature, and return later to see any changes.
The application utilizes image-analysis techniques, including Excess Green (ExG), Edge Correlation, and Framing Assessment. ExG measures green-channel dominance in an image, Sobel edge information and structural correlation assess whether two photographs are sufficiently comparable, and pixel-level differences are calculated. However, large pixel differences are not automatically evidence of biological change, and the system may not have enough evidence to make claims.
The scientific reliability gate ensures that the application does not confidently claim directional change when visual evidence is unreliable. WildTrace's nature-inspired interface brings together habitat observations, comparisons, and field missions, providing a comprehensive experience for users. Unlike image identification tools, WildTrace is designed for revisiting a place and building an observation history.
WildTrace combines several image-analysis techniques to characterize differences between observations. The application uses OpenRouter-based multimodal inference, configurable vision models, Zod validation of model responses, bounded retries, and configurable timeouts. A deterministic local heuristic fallback is utilized when hosted inference is unavailable.
The app uses Next.js and TypeScript for the web application and API routes, Prisma and SQLite for persistence, Sharp for image processing, computer vision utilities for Excess Green calculation, Sobel edge analysis, and pixel-difference measurements, OpenRouter for hosted multimodal inference, and Zod for structured response validation. The application is containerized using Docker Compose, with image processing, AI inference, structured validation, and fallback behavior separated for modularity.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
