Fore-Mamba3D: Mamba-based Foreground-Enhanced Encoding for 3D Object Detection

2026-02-23

AI Summary

This work addresses the limitations of existing Mamba-based 3D object detection methods, which either suffer from excessive background noise when processing full-scene voxel sequences or experience response decay and contextual loss when encoding only foreground voxels. To overcome these challenges, we propose Fore-Mamba3D, the first approach to enable efficient foreground voxel encoding within the Mamba framework. Our method employs high-confidence foreground sampling to focus on critical regions, introduces a Region-to-Global Sliding Window (RGSW) mechanism to mitigate cross-instance response decay, and incorporates a Semantic-Assisted State Space Fusion module (SASFMamba) to enhance both geometric and semantic awareness. Extensive experiments demonstrate that Fore-Mamba3D achieves state-of-the-art performance across multiple 3D detection benchmarks, validating the effectiveness and generalizability of our foreground-enhanced strategy.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Sampling/Simulation-based Search
• Intelligent Robots: State Estimation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Linear modeling methods like Mamba have been merged as the effective backbone for the 3D object detection task. However, previous Mamba-based methods utilize the bidirectional encoding for the whole non-empty voxel sequence, which contains abundant useless background information in the scenes. Though directly encoding foreground voxels appears to be a plausible solution, it tends to degrade detection performance. We attribute this to the response attenuation and restricted context representation in the linear modeling for fore-only sequences. To address this problem, we propose a novel backbone, termed Fore-Mamba3D, to focus on the foreground enhancement by modifying Mamba-based encoder. The foreground voxels are first sampled according to the predicted scores. Considering the response attenuation existing in the interaction of foreground voxels across different instances, we design a regional-to-global slide window (RGSW) to propagate the information from regional split to the entire sequence. Furthermore, a semantic-assisted and state spatial fusion module (SASFMamba) is proposed to enrich contextual representation by enhancing semantic and geometric awareness within the Mamba model. Our method emphasizes foreground-only encoding and alleviates the distance-based and causal dependencies in the linear autoregression model. The superior performance across various benchmarks demonstrates the effectiveness of Fore-Mamba3D in the 3D object detection task.

Problem

• 3D object detection
• Mamba
• foreground encoding
• response attenuation
• context representation

Innovation

• Mamba
• foreground-enhanced encoding
• 3D object detection
• regional-to-global slide window
• semantic-assisted fusion