1
Building Food Metadata with LLM Juries
DoorDash用LLM陪审团与上下文优化，多模态AI精准构建食品元数据，工程实践亮点。
Article URL: https://careersatdoordash.com/blog/building-food-metadata-with-llm-juries-context-optimization-multimodal-ai/ Comments URL: https://news.…
2
VisReason: A Large-Scale Dataset for Visual Chain-of-Thought Reasoning
VisReason数据集带来视觉推理新突破，大规模多步骤链式思维数据推动多模态AI发展。
arXiv:2511.17731v2 Announce Type: replace-cross Abstract: Chain-of-Thought (CoT) prompting has proven remarkably effective for eliciting complex reaso…
3
BrepLLM: Enabling Large Language Models to Understand Boundary Representations
让大语言模型首次理解3D CAD边界表示，打破自然语言与几何建模的壁垒。
arXiv:2512.16413v2 Announce Type: replace Abstract: Current token-sequence-based Large Language Models (LLMs) struggle to directly process 3D Boundary…
4
DMV-Bench: Diagnosing Long-Horizon Multimodal Agents' Visual Memory with Incidental Cue Injection
多模态AI能记住多久？DMV-Bench新基准用偶然线索注入诊断视觉记忆缺陷，揭示智能体长周期任务漏洞。
arXiv:2606.27499v1 Announce Type: cross Abstract: Research on agent memory has matured rapidly, but almost entirely on the text side: few existing ben…
5
MEDLAYXPLAIN: Benchmarking the Expert-Lay Gap in Medical Vision-Language Models
新基准MEDLAYXPLAIN系统评估了33个医学视觉语言模型，揭示专家和外行在理解上的显著认知差距。
arXiv:2606.21194v1 Announce Type: cross Abstract: Medical Vision-Language Models (Med-VLMs) achieve strong expert-level performance, yet their ability…
6
Multimodal Evaluator Preference Collapse: Cross-Modal Contagion in Self-Evolving Agents
多模态AI自我评估存在系统性偏差，一种新发现的“偏好崩溃”现象加剧了模型反馈循环中的错误。
arXiv:2606.16682v1 Announce Type: new Abstract: When AI agents use language models to evaluate their own outputs in a feedback loop, systematic biases…
7
Sustainability assessment using multimodal AI agents
多模态AI智能体协同工作，让可持续性评估更高效精准，这是AI落地环保的新思路。
arXiv:2507.17012v2 Announce Type: replace Abstract: Reducing the rapidly growing environmental impact of the computing industry requires assessing the…
8
MSUE: Multi-Modal Soccer Understanding Expert
融合视觉与文本数据，打造专业足球AI分析师，新论文发布多模态足球理解专家MSUE。
arXiv:2606.12106v1 Announce Type: cross Abstract: This paper presents our solution to the 2026 SoccerNet VQA Challenge. We first develop a cost-effect…
9
Imagination Helps Visual Reasoning, But Not Yet in Latent Space
探讨想象力如何辅助视觉推理，并揭示其在隐空间中的局限性，为多模态AI研究提供新视角。
arXiv:2602.22766v2 Announce Type: replace Abstract: Latent visual reasoning aims to mimic human's imagination process by meditating through hidden sta…
10
Noise-Aware Visual Representation Learning for Medical Visual Question Answering
医学视觉问答新突破：噪声感知学习提升视觉表示鲁棒性
arXiv:2606.05535v1 Announce Type: cross Abstract: Medical visual question answering (Med-VQA) has strong potential for clinical decision support by en…
11
Google Gemma 4 12B
Google 最新多模态模型 Gemma 4 12B 无需编码器即可本地运行，开源灵活且高效。
Run multimodal AI locally with an encoder-free architecture Discussion | Link
12
SSR3D-LLM: Structured Spatial Reasoning via Latent Steps for Fine-Grained Grounding in Unified 3D-LLMs
新提出的SSR3D-LLM框架通过隐式步骤实现结构化空间推理，让3D大模型具备细粒度定位能力，解决统一3D-LLM中的空间理解难题。
arXiv:2605.28490v1 Announce Type: cross Abstract: 3D object grounding localizes referred objects in a 3D scene from natural language. Unified instance…
13
Stop Describing Your Bugs — Just Screenshot Them
截图自动分析bug，无需长篇描述，Gemma 4多模态让问题定位一步到位，效率拉满！
This is a submission for the Gemma 4 Challenge: Build with Gemma 4 What I Built BugLens is an AI-powered screenshot debugger for developers. You uploa…
14
How Netflix is Using Multimodal AI to Power Video Search
Netflix用多模态AI将216百万帧原始画面转化为数十亿数据点，颠覆视频搜索体验。
In this article, we will understand how Netflix built this system and the challenges it faced.