Pytorch-Wildlife: A Collaborative Deep Learning Framework for Conservation

TL;DR

The paper tackles the bottleneck of processing large camera-trap datasets for biodiversity monitoring by introducing Pytorch-Wildlife, an open-source, PyTorch-based framework designed around accessibility, scalability, and transparency. Its modular architecture comprises datasets, a model zoo, and utilities, with integrations to the LILA-BC ecosystem and a user-friendly interface that supports local and remote deployment. The work yields practical contributions including the MegaDetectorV6-compact detector, a classification fine-tuning module, and a community-driven leaderboard to guide model selection, demonstrated through two real-world deployments in the Amazon and Galápagos. The framework enables conservation researchers to deploy robust AI workflows on varied hardware with human-in-the-loop capabilities while upholding openness and privacy safeguards, facilitating broader adoption and future task expansion across conservation domains.

Abstract

The alarming decline in global biodiversity, driven by various factors, underscores the urgent need for large-scale wildlife monitoring. In response, scientists have turned to automated deep learning methods for data processing in wildlife monitoring. However, applying these advanced methods in real-world scenarios is challenging due to their complexity and the need for specialized knowledge, primarily because of technical challenges and interdisciplinary barriers. To address these challenges, we introduce Pytorch-Wildlife, an open-source deep learning platform built on PyTorch. It is designed for creating, modifying, and sharing powerful AI models. This platform emphasizes usability and accessibility, making it accessible to individuals with limited or no technical background. It also offers a modular codebase to simplify feature expansion and further development. Pytorch-Wildlife offers an intuitive, user-friendly interface, accessible through local installation or Hugging Face, for animal detection and classification in images and videos. As two real-world applications, Pytorch-Wildlife has been utilized to train animal classification models for species recognition in the Amazon Rainforest and for invasive opossum recognition in the Galapagos Islands. The Opossum model achieves 98% accuracy, and the Amazon model has 92% recognition accuracy for 36 animals in 90% of the data. As Pytorch-Wildlife evolves, we aim to integrate more conservation tasks, addressing various environmental challenges. Pytorch-Wildlife is available at https://github.com/microsoft/CameraTraps.

Figures (2)

• Figure 1: Overview figure of the Pytorch-wildlife framework. First, it connects to the datasets available on LILA-BC for ease of training and validation; second, it offers detection and classification models with pretrained weights in a variety of datasets; finally, it comes with a user interface and a set of utility functions for visualization and adequate post-processing.
• Figure 2: Pytorch-wildlife user interface. It allows the user to load detection and classification models, as well as to perform single image detection, batch image detection, and single video detection with a confidence threshold for human validation.