# Toward Faithful Segmentation Attribution via Benchmarking and Dual-Evidence Fusion

Source: [Toward Faithful Segmentation Attribution via Benchmarking and Dual-Evidence Fusion](<https://www.sota2.com/research/paper/69c36bfc0636e5bf4afc6425>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-26T05:00:04.122Z

Authors: Abu Noman Md Sakib, OFM Riaz Rahman Aranya, Kevin Desai, Zijie Zhang

Published: 2026-03-23T22:52:00.000Z

## Abstract

Attribution maps for semantic segmentation are almost always judged by visual plausibility. Yet looking convincing does not guarantee that the highlighted pixels actually drive the model's prediction, nor that attribution credit stays within the target region. These questions require a dedicated evaluation protocol. We introduce a reproducible benchmark that tests intervention-based faithfulness, off-target leakage, perturbation robustness, and runtime on Pascal VOC and SBD across three pretrained backbones. To further demonstrate the benchmark, we propose Dual-Evidence Attribution (DEA), a lightweight correction that fuses gradient evidence with region-level intervention signals through agreement-weighted fusion. DEA increases emphasis where both sources agree and retains causal support when gradient responses are unstable. Across all completed runs, DEA consistently improves deletion-based faithfulness over gradient-only baselines and preserves strong robustness, at the cost of additional compute from intervention passes. The benchmark exposes a faithfulness-stability tradeoff among attribution families that is entirely hidden under visual evaluation, providing a foundation for principled method selection in segmentation explainability. Code is available at https://github.com/anmspro/DEA.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Attribution Analysis on SBD](<https://www.sota2.com/research/sota/attribution-analysis-on-sbd.md>): Attribution Analysis; SBD; TDD: 38.1.

- [Attribution Analysis on VOC](<https://www.sota2.com/research/sota/attribution-analysis-on-voc.md>): Attribution Analysis; VOC; TDD: 44.9.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2603.22624>)

[PDF](<https://arxiv.org/pdf/2603.22624.pdf>)
