How This Fulbright Scholar is Using Computer Vision to Protect Endangered Species

Kasim Rafiq, a conservationist and wildlife biologist, has developed a novel approach using computer vision and edge machine learning to enhance wildlife surveys, particularly for African predators, by leveraging tourist photographs to reduce costs significantly compared to traditional methods. This innovative method, supported by a rapid prototyping grant from ConservationX, involves the use of vehicle-mounted cameras on safari game drives to automatically detect and photograph wildlife, thus facilitating continuous and rapid species assessments. Rafiq's work, which builds on a rich dataset of leopard images gathered through collaborations with organizations like WildMe and Botswana Predator Conservation, utilizes Google Cloud's AutoML Vision Edge models for real-time inference and Roboflow for data preparation and augmentation. The prototype, initially deployed using a Raspberry Pi setup, aims to expand its scope to detect a broader range of wildlife and improve hardware resilience for use in the African bush, with Rafiq inviting collaboration with technology enthusiasts to further advance the project.