Cell tracking with multifeature fusion

Abstract Cell tracking is currently a powerful tool in variety of biomedical research topics. Most cell algorithms follow the by detection paradigm. Detection critical for subsequent tracking. Unfortunately, very accurate not easy due to many factors like densely populated, low contrast, and possible impurities included. Keeping multiple cells across frames suffers difficulties, as may have similar appearance, they change their shapes, nearby interact each other. In this paper, we propose unified tracking-by-detection framework, where detector AttentionUnet++, multimodal extension Efficient Convolution Operators algorithm, an effective data association algorithm are Experiments show that proposed can outperform existing algorithms.