✨SlowBA: An efficiency backdoor attack towards VLM-based GUI agents

📝 Summary:
SlowBA is a novel backdoor attack targeting the response latency of VLM-based GUI agents. It induces excessively long reasoning chains using realistic pop-up window triggers, significantly increasing response length and latency while maintaining task accuracy. This reveals a new security vulnerab...

🔹 Publication Date: Published on Mar 9

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.08316
• PDF: https://arxiv.org/pdf/2603.08316
• Github: https://github.com/tu-tuing/SlowBA

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#BackdoorAttack #AISecurity #VLM #GUIagents #Cybersecurity

✨NanoVDR: Distilling a 2B Vision-Language Retriever into a 70M Text-Only Encoder for Visual Document Retrieval

📝 Summary:
NanoVDR improves visual document retrieval by distilling a large VLM teacher into a small 70M text-only query encoder. This decouples document indexing from query processing, achieving 50x lower latency and 32x fewer parameters with nearly identical quality.

🔹 Publication Date: Published on Mar 13

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.12824
• PDF: https://arxiv.org/pdf/2603.12824
• Project Page: https://huggingface.co/nanovdr

🔹 Models citing this paper:
• https://huggingface.co/nanovdr/NanoVDR-L
• https://huggingface.co/nanovdr/NanoVDR-S-Multi
• https://huggingface.co/nanovdr/NanoVDR-S

✨ Spaces citing this paper:
• https://huggingface.co/spaces/nanovdr/NanoVDR-Demo

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VisualDocumentRetrieval #ModelDistillation #VLM #InformationRetrieval #DeepLearning

✨Anatomy of a Lie: A Multi-Stage Diagnostic Framework for Tracing Hallucinations in Vision-Language Models

📝 Summary:
Vision-Language Models (VLMs) frequently "hallucinate" - generate plausible yet factually incorrect statements - posing a critical barrier to their trustworthy deployment. In this work, we propose a n...

🔹 Publication Date: Published on Mar 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.15557
• PDF: https://arxiv.org/pdf/2603.15557
• Github: https://github.com/Lexiang-Xiong/CAD

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #AIHallucinations #TrustworthyAI #ExplainableAI #AIResearch

✨TerraScope: Pixel-Grounded Visual Reasoning for Earth Observation

📝 Summary:
TerraScope is a new VLM for Earth Observation enabling pixel-grounded geospatial reasoning. It offers modality-flexible and multi-temporal capabilities, outperforming existing models on a new benchmark for accurate and interpretable results.

🔹 Publication Date: Published on Mar 19

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.19039
• PDF: https://arxiv.org/pdf/2603.19039
• Project Page: https://shuyansy.github.io/terrascope/
• Github: https://github.com/shuyansy/Earth-Observation-VLMs

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#EarthObservation #VLM #Geospatial #RemoteSensing #ComputerVision

✨From Masks to Pixels and Meaning: A New Taxonomy, Benchmark, and Metrics for VLM Image Tampering

📝 Summary:
This paper shifts VLM image tampering detection from coarse object masks to pixel-level analysis with semantic understanding. It introduces a new taxonomy, benchmark, and metrics to evaluate both localization accuracy and the meaning of image modifications. This offers a more rigorous standard fo...

🔹 Publication Date: Published on Mar 20

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.20193
• PDF: https://arxiv.org/pdf/2603.20193
• Github: https://github.com/VILA-Lab/PIXAR

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #ImageTampering #DeepfakeDetection #ComputerVision #AIResearch

✨ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Model

📝 Summary:
ThinkJEPA improves latent world models by combining dense JEPA dynamics with VLM semantic guidance through a dual-temporal pathway. This framework enhances long-horizon hand-manipulation trajectory prediction.

🔹 Publication Date: Published on Mar 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.22281
• PDF: https://arxiv.org/pdf/2603.22281

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#ThinkJEPA #LatentWorldModels #VLM #Robotics #AI

✨RealChart2Code: Advancing Chart-to-Code Generation with Real Data and Multi-Task Evaluation

📝 Summary:
RealChart2Code is a new benchmark assessing VLM ability to generate complex, multi-panel charts from real data. It reveals significant performance gaps between proprietary and open-weight models, highlighting VLM struggles with intricate plots.

🔹 Publication Date: Published on Mar 26

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.25804
• PDF: https://arxiv.org/pdf/2603.25804
• Project Page: https://huggingface.co/datasets/zjj1233/RealChart2Code
• Github: https://github.com/Speakn0w/RealChart2Code

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #ChartToCode #Benchmark #AI #DataScience

✨A Matter of Time: Revealing the Structure of Time in Vision-Language Models

📝 Summary:
This paper reveals that vision-language models embed temporal information in a structured way. It introduces a new dataset and methods to derive explicit timeline representations from these models, enabling efficient temporal reasoning.

🔹 Publication Date: Published on Oct 22, 2025

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.19559
• PDF: https://arxiv.org/pdf/2510.19559
• Project Page: https://tekayanidham.github.io/timeline-page/
• Github: https://github.com/TekayaNidham/timeline-vlm

✨ Spaces citing this paper:
• https://huggingface.co/spaces/Nidhamtek/timeline-vlm

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #TemporalReasoning #AIResearch #MachineLearning #DeepLearning

✨Memory-Augmented Vision-Language Agents for Persistent and Semantically Consistent Object Captioning

📝 Summary:
A memory-augmented VLM agent resolves inconsistent object descriptions across viewpoints. It unifies data association, captioning, and exploration within a single framework, leveraging object-level memory for persistent semantic consistency and improved scores.

🔹 Publication Date: Published on Mar 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.24257
• PDF: https://arxiv.org/pdf/2603.24257
• Project Page: https://hsp-iit.github.io/epos-vlm/

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #ObjectCaptioning #AI #ComputerVision #DeepLearning

✨VisionFoundry: Teaching VLMs Visual Perception with Synthetic Images

📝 Summary:
VisionFoundry creates synthetic visual question answering data using LLMs and text-to-image models to improve VLM visual perception. Training with this targeted data significantly boosts model performance on visual perception benchmarks like MMVP and CV-Bench-3D.

🔹 Publication Date: Published on Apr 10

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.09531
• PDF: https://arxiv.org/pdf/2604.09531
• Project Page: https://zlab-princeton.github.io/VisionFoundry/

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VLM #VisualPerception #SyntheticData #LLM #AI