Bounding Box Tightness Prior for Weakly Supervised Image Segmentation
This work addresses image segmentation for medical applications with weak supervision, offering an incremental improvement over existing methods.
The paper tackles weakly supervised image segmentation using bounding box annotations, proposing a method that integrates a bounding box tightness prior via generalized multiple instance learning and smooth maximum approximation. It achieves state-of-the-art performance on two medical datasets, as measured by Dice coefficient.
This paper presents a weakly supervised image segmentation method that adopts tight bounding box annotations. It proposes generalized multiple instance learning (MIL) and smooth maximum approximation to integrate the bounding box tightness prior into the deep neural network in an end-to-end manner. In generalized MIL, positive bags are defined by parallel crossing lines with a set of different angles, and negative bags are defined as individual pixels outside of any bounding boxes. Two variants of smooth maximum approximation, i.e., $α$-softmax function and $α$-quasimax function, are exploited to conquer the numeral instability introduced by maximum function of bag prediction. The proposed approach was evaluated on two pubic medical datasets using Dice coefficient. The results demonstrate that it outperforms the state-of-the-art methods. The codes are available at \url{https://github.com/wangjuan313/wsis-boundingbox}.