Search by image in camera trap dataset

Akin Kaki


Supervised by Charith Perera; Moderated by Carolina Fuentes Toro

This project focuses on designing a technique that allows researchers to search for images in the same location in a camera trap dataset by using few images from that location or using a single image when the whole dataset is processed using the technique. Using the same principle, the presence of animal in the camera trap image is also identified. and finally by using the results from the above, this project also focus on how normalization of data can affect the perfomance of a machine learning model when trained using camera trap dataset when training to identify animal present in an image. This is done as continuation to my placement project, where i have developed an ensemble model to identify the presence of animal in an image and front-end it.

Final Report (23/09/2022) [Zip Archive]

Publication Form