Audio Pattern Ranger (https://github.com/audio-pattern-ranger/apr) is a monitoring framework using custom-trained models to detect local disturbances. This works by letting a user review videos that are broken into 1-second clips and using that data to train a model which is then used to evaluate each 1-second clip in longer recordings. The current version can produce models that are quite accurate, but the number of false positives is quite problematic, especially because most false positives observed are caused by neutral background noise. The goal of this project is to resolve false positives so that we can have the appearance of 100% accuracy
Years: Any
Location: Africa
Requested on: 2024-09-30
Deep learning, PyTorch, Tensorflow, Open AI