Improving In-Field Cassava Whitefly Pest Surveillance With Machine Learning

Jeremy Francis Tusubira, Solomon Nsumba, Flavia Ninsiima, Benjamin Akera, Guy Acellam, Joyce Nakatumba, Ernest Mwebaze, John Quinn, Tonny Oyana; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2020, pp. 68-69

Whiteflies are the major vector responsible for the transmission of cassava related diseases in tropical environments, and knowing the numbers of whiteflies is key in detecting and identifying their spread and prevention. However, the current approach for counting whiteflies is a simple visual inspection, where a cassava leaf is turned upside down to reveal the underside where the whiteflies reside to enable a manual count. Repeated across many cassava farms, this task is quite tedious and time-consuming. In this paper, we propose a method to automatically count whiteflies using computer vision techniques. To implement this approach, we collected images of infested cassava leaves and trained a computer vision detector using Haar Cascade and DeepLearning techniques. The two techniques were used to identify the pest in images and return a count. Our results show that this novel method produces a whitefly count with high precision. This method could be applied to similar object detection scenarios similar to the whitefly problem with minor adjustments.