Simple Does It: Weakly Supervised Instance and Semantic Segmentation

TL;DR

This paper introduces a weakly supervised method for semantic and instance segmentation that uses bounding box annotations, achieving nearly full performance of fully supervised models with minimal modifications.

Contribution

The authors propose a novel approach that leverages bounding box annotations for segmentation tasks without altering the training procedure, significantly improving weakly supervised results.

Findings

Achieves ~95% of fully supervised segmentation quality.

Requires only a single training round.

Does not modify segmentation training procedures.

Abstract

Semantic labelling and instance segmentation are two tasks that require particularly costly annotations. Starting from weak supervision in the form of bounding box detection annotations, we propose a new approach that does not require modification of the segmentation training procedure. We show that when carefully designing the input labels from given bounding boxes, even a single round of training is enough to improve over previously reported weakly supervised results. Overall, our weak supervision approach reaches ~95% of the quality of the fully supervised model, both for semantic labelling and instance segmentation.