This was my first ever computer vision/machine learning project and first ever programming experience in Python. I did this project as a visiting researcher at the University of Guelph under the supervision of Dr. Graham Taylor. We did it in collaboration with SemiosBio Inc. and the School of Enviromental Sciences.
The goal of this project was to recognize diamond-back moths from a variety of insects using computer vision and machine learning. The dataset consisted of images taken from a real insect trapping device that trapped all kinds of insects including moths.
There were two main steps in this project:
- Develop a image annotation/labelling GUI that would enable our colleagues from the School of Environmental Sciences to manually label the diamond-back moths in these images and generate metadata about the state of each label (e.g. truncated, occluded, blurry, good, bad etc.).
- Develop a insect recognition system that would be able to recognize the diamond-back moths in every image.
Below is a sample image with some moths identified by the model:
