In this paper we describe a scheme for automatic identification of wood logging activity in forest based on acoustic surveillance. Specifically, we evaluate five machine learning classification algorithms using several audio descriptors for the identification of chainsaw wood logging sounds in the noisy environment of a forest. Different environmental noise interference levels, in terms of sound-to-noise ratio, were considered and the best performance was achieved using support vector machines.
|ACM International Conference Proceeding Series
|9th Hellenic Conference on Artificial Intelligence, SETN 2016
|18/05/16 → 20/05/16
- Audio based surveillance
- Audio processing
- Biodiversity monitoring