Wheat Head Detection using Deep, Semi-Supervised and Ensemble Learning

摘要

In this paper, we propose an object detection methodology applied to Global Wheat HeadDetection (GWHD) Dataset. We have been through two major architectures of object detectionwhich are Faster R-CNN, and EfficientDet, in order to design a novel and robust wheathead detection model. We emphasize on optimizing the performance of our proposed finalarchitectures. Furthermore, we have been through an extensive exploratory data analysis,data cleaning, data splitting and adapted best data augmentation techniques to our context.We use semi supervised learning, precisely pseudo-labeling, to boost previous supervisedmodels of object detection. Moreover, we put much effort on ensemble learningincluding test time augmentation, multi-scale ensemble and bootstrap aggregating toachieve higher performance. Finally, we use weighted boxes fusion as our post processingtechnique to optimize our wheat head detection results. Our solution has been submittedto solve a research challenge launched on the GWHD Dataset which was led by nineresearch institutes from seven countries. Our proposed method was ranked within the top6% in the above-mentioned challenge.