Challenge Overview

Camera traps used in wildlife monitoring capture a massive number of images - typically 30,000 to 40,000 within a 15-day survey. Due to high sensitivity, most of these images are triggered by non-relevant movements such as leaves, shadows, or environmental noise, making manual sorting highly time-consuming and resource-intensive.

Challenge Objective

Develop an AI/ML-based system that can automatically classify camera trap images into:

• Images containing wildlife
• Images without wildlife

Problem Scope

• Handling large-scale image datasets from camera traps
• Filtering irrelevant images caused by false triggers
• Accurately detecting wildlife presence

Key Issues

• High volume of noisy and irrelevant data
• Time-consuming manual image analysis
• Difficulty in detecting animals in low light or occluded conditions
• Need for scalable and efficient processing

Guiding Questions

• How can AI models accurately distinguish animals from background noise?
• What techniques improve detection in challenging environments (night, motion blur)?
• How can processing be optimized for large datasets?
• Can the model be deployed on edge devices for faster filtering?