AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

CSI-TextBench: A Dataset and Benchmark for Language-Grounded Ambient Sensing Perception

Guozhen Zhu, Yuqian Hu, Sakila S Jayaweera, Wei-Hsiang Wang, Beibei Wang, K. Liu

NeurIPS 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Multimodal and affective interaction modeling · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Introduces CSI-TextBench, a dataset and benchmark pairing WiFi channel state information with language for language-grounded indoor ambient sensing perception.

Problem

Ambient WiFi CSI captures indoor human activity and environmental context passively, but it remains weakly connected to language-aligned multimodal systems.

Approach

A benchmark that frames WiFi CSI perception as language-grounded tasks, pairing CSI signals with text so that language-aligned models can be evaluated on indoor sensing.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use contrastive CLIP InfoNCE instead of Barlow Twins for CSI-text alignmentCSI-text dual encoder · Fall detection zero-shotaccuracy (%): 59.8 -> 94.2Improved
use CLIP InfoNCE instead of SigLIP on 276-class SignFiCSI-text dual encoder · SignFi zero-shotaccuracy (%): 71.8 -> 97.0Improved
contrastive CSI-text alignment versus supervised classifier with same encoderCLIP InfoNCE · HAR zero-shotaccuracy (%): 97.9 -> 97.8No clear effect
swap ResNet-18 1D CSI encoder for ViT-1D or PatchTST (ablation)CLIP InfoNCE on HAR · HAR zero-shotaccuracy (%): 98.7% -> 95.5% (ViT-1D), 94.1% (PatchTST)Hurt
swap frozen CLIP text encoder for MiniLM-L6-v2 (ablation)CLIP InfoNCE on HAR · HAR zero-shotaccuracy (%): 98.7% -> 90.2%Hurt
add Level 2 compositional captions to Level 1 for single-task training (ablation)CLIP InfoNCE · HAR and Proximity zero-shotaccuracy (%): difference at most 0.3 pointsNo clear effect
anchor expert-language prompts in a natural sentence frame (ablation)template-trained CLIP InfoNCE · HAR open-vocabularyaccuracy (%): 59.1 -> 96.2Improved
anchor sensor-grounded prompts in a natural sentence frame (ablation)template-trained CLIP InfoNCE · HAR open-vocabularyaccuracy (%): 35.4 -> 77.1Improved
train on human-written free-form text instead of templates (ablation)CLIP InfoNCE on HAR · HAR zero-shot, specialized promptsaccuracy (%): 59.1 -> 87.0 (expert raw), 35.4 -> 80.2 (sensor raw)Improved
train on human-written free-form text instead of templates (ablation)CLIP InfoNCE on HAR · HAR zero-shot, ordinary promptsaccuracy (%): 95.9 -> 94.5 (train template), 96.0 -> 93.9 (layperson)Hurt
joint multitask training on all 7 tasks with shared encoderCLIP InfoNCE · Proximity zero-shotaccuracy (%): 94.6 -> 84.5Hurt
retrieve composite scene descriptions jointly instead of independent per-attribute QACLIP InfoNCE multitask model · Composite scenes (140)joint/top-1 accuracy (%): 70.4 -> 86.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Experimental setup

Benchmarks: CSI-TextBench

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy