Hi, welcome to FoxFinder! The model is designed to identify foxes and other vulnerable wildlife that are being hunted illegally by fox hunting packs in the UK. Hunt saboteurs use drones to monitor for illegal hunting, but it is extremely difficult to spot foxes both in real time on the small screen of a drone remote, and also in reviewing of footage. This model's primary goal is to allow for saboteurs to find foxes hidden in 5+ hours of footage that may be taken over the course of a day of hunting as fast and as easily as possible, in order to assist police in investigating and prosecuting. The model's secondary goal is to enable real-time analysis of footage on location. This could happen either by processing of video output from the drone's remote, or by integrating the model directly into the drone's firmware, which has recently become possible in DJI's industrial drones.
This is currently a project I am undertaking single-handedly, with saboteur groups providing valuable footage of foxes.
The model has already successfully detected foxes in untrained footage, and where the fox was not spotted by human observer either during the moment, or reviewing the footage, so a very promising start. This model will almost certainly provide a massive number of false-positives, so human analysis of model output remains essential.
Model tree for Bfreak1/FoxFinder
Base model
Ultralytics/YOLO11