One of our goals is to classify fish automatically by AI. Since the labeling work for fish classification requires expertise and labour, we looked for existing dataset to test the algorithms. On a group meeting in Jan, 2017, my colleague Peng Sun recommended Kaggle, a platform for predictive modelling and analytics competitions. He also stated he had seen Read more
Sponsored by Missouri Department of Conservation (MDC), we are working on fish recognition and measurement based on computer vision. The problem: MDC fisheries biologists measure fish in Missouri waterways to get an understanding of the fish species, the waterway and the health of both. This can be a time-consuming task that involves net catch-and-release of multiple fish and Read more
We developed a software for vegetation density estimation, which automatically detects obscured region of reference object (typically a blackboard) covered by grass under different illumination conditions in the wild. Reference object and grass are firstly distinguished using a LBP+SVM classifier. And then a K-Means color model is used to refine the prediction at pixel level. Read more
The demo video above shows my previous work done when I was a member of Computer Vision Lab at University of Missouri in the year of 2013. We proposed a new feature (Normal Vector Pattern) to describe the surface of an object in raw depth images, significantly outperformed current state-of-the-art depth image features with a Read more
We implemented an indoor positioning system for the Naka Hall (old name Engineering Building West) at University of Missouri. With a photo captured (by cell phone) inside the building used as the input image, the system can show the location where the photo is most likely to be captured. 3 meters precision and 86% accuracy Read more
