Recursively Refined R-CNN: Instance Segmentation with Self-RoI Rebalancing

TL;DR

The paper introduces R^3-CNN, a recursive instance segmentation model that improves accuracy and reduces complexity by replacing cascade architectures with a loop mechanism and recursive re-sampling, outperforming recent models.

Contribution

It proposes a novel recursive re-sampling and loop mechanism in R^3-CNN that enhances instance segmentation performance while reducing model complexity.

Findings

R^3-CNN surpasses HTC in accuracy on COCO dataset.

It achieves performance improvements independently of the baseline model.

The architecture significantly reduces the number of parameters.

Abstract

Within the field of instance segmentation, most of the state-of-the-art deep learning networks rely nowadays on cascade architectures, where multiple object detectors are trained sequentially, re-sampling the ground truth at each step. This offers a solution to the problem of exponentially vanishing positive samples. However, it also translates into an increase in network complexity in terms of the number of parameters. To address this issue, we propose Recursively Refined R-CNN (R^3-CNN) which avoids duplicates by introducing a loop mechanism instead. At the same time, it achieves a quality boost using a recursive re-sampling technique, where a specific IoU quality is utilized in each recursion to eventually equally cover the positive spectrum. Our experiments highlight the specific encoding of the loop mechanism in the weights, requiring its usage at inference time. The R^3-CNN architecture is able to surpass the recently proposed HTC model, while reducing the number of parameters significantly. Experiments on COCO minival 2017 dataset show performance boost independently from the utilized baseline model. The code is available online at https://github.com/IMPLabUniPr/mmdetection/tree/r3_cnn.