Improving Pixel Embedding Learning through Intermediate Distance Regression Supervision for Instance Segmentation
This work addresses instance segmentation for computer vision applications, offering an incremental improvement over existing methods.
The paper tackles instance segmentation by proposing a method that incorporates a distance regression module to improve pixel embedding learning, resulting in an over 8% improvement in mSBD scores on the CVPPP Leaf Segmentation Challenge and achieving the best overall result on the leaderboard.
As a proposal-free approach, instance segmentation through pixel embedding learning and clustering is gaining more emphasis. Compared with bounding box refinement approaches, such as Mask R-CNN, it has potential advantages in handling complex shapes and dense objects. In this work, we propose a simple, yet highly effective, architecture for object-aware embedding learning. A distance regression module is incorporated into our architecture to generate seeds for fast clustering. At the same time, we show that the features learned by the distance regression module are able to promote the accuracy of learned object-aware embeddings significantly. By simply concatenating features of the distance regression module to the images as inputs of the embedding module, the mSBD scores on the CVPPP Leaf Segmentation Challenge can be further improved by more than 8% compared to the identical set-up without concatenation, yielding the best overall result amongst the leaderboard at CodaLab.