aiqbench analysis
This research addresses the critical constraint of deploying high-level visual affordance models on resource-constrained wearable robots by optimizing the decoder module. The proposed lightweight neural networks successfully balance generalization performance with low compute costs, outperforming existing baselines in real-world datasets without sacrificing accuracy for edge deployment scenarios.
Lightweight Neural Networks for Affordance Segmentation: Enhancement of the Decoder Module
arXiv:2607.29473v1 Announce Type: cross Abstract: The deployment of deep neural networks for visual affordance segmentation on wearable robots poses may prove critical, due to some conflicting aspects of the problem. On one hand, affordance segmentation requires high-level abstraction capabilities, that typically involve large-size models. On the other hand, computing resources hosted on wearable robots prevent to run large-size models in real-time. The paper presents an analysis of the role of the segmentation head in the trade-off between generalization performance and compute cost. The obtained models outperform modern baseline solutions in well-known, real-world datasets while meeting low computing requirements.
Abstract:The deployment of deep neural networks for visual affordance segmentation on wearable robots poses may prove critical, due to some conflicting aspects of the problem. On one hand, affordance segmentation requires high-level abstraction capabilities, that typically involve large-size models. On the other hand, computing resources hosted on wearable robots prevent to run large-size models in real-time. The paper presents an analysis of the role of the segmentation head in the trade-off between generalization performance and compute cost. The obtained models outperform modern baseline solutions in well-known, real-world datasets while meeting low computing requirements.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Performance (cs.PF)
arXiv:2607.29473 [cs.CV]
(or arXiv:2607.29473v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2607.29473 arXiv-issued DOI via DataCite (pending registration)
S. Lugani, E. Ragusa, R. Zunino, and P. Gastaldo, "Lightweight Neural Networks for Affordance Segmentation: Enhancement of the Decoder Module" in Applications in Electronics Pervading Industry, Environment and Society. ApplePies 2023
https://doi.org/10.1007/978-3-031-48121-5_63 DOI(s) linking to related resources
Full text under the publisher's open licence · 156 words
Related coverage
News
Quantum Computing Report · Jul 25, 2026
Feature Reconfiguration With Visual Prior for Medical Lesion Segmentation
arXiv · Sep 5, 2026
LiteViLNet: Lightweight Vision-LiDAR Fusion Network for Efficient Road Segmentation
arXiv · Sep 17, 2026
The Machines that Make the Machines
NVIDIA · Oct 7, 2026