This project aims to detect wheat heads in an image with YOLONAS. The dataset is Global Wheat Detection. The dataset has pictures of wheat heads from farms worldwide. Each picture is big, at 1024x1024 pixels. We have many pictures for learning, over 3300, and some for testing, around 1000. A tricky part is that some pictures don't show wheat heads, while others have up to 116. A sample of images from this dataset and corresponding labels is shown in the figure below.
YOLONAS is used to detect these bounding boxes. The output of test images from this model is shown in the figure below.
[1] Etaati, Sepideh, Javad Khoramdel, and Esmaeil Najafi. "Automated Wheat Disease Detection using Deep Learning: an Object Detection and Classification Approach." In 2023 11th RSI International Conference on Robotics and Mechatronics (ICRoM), pp. 116-121. IEEE, 2023.