video-qa

[EMNLP 2023] TESTA: Temporal-Spatial Token Aggregation for Long-form Video-Language Understanding

[ICLR2024] Codes and Models for COSA: Concatenated Sample Pretrained Vision-Language Foundation Model

Unifying the Video and Question Attentions for Open-Ended Video Question Answering

Video Question Answering via Hierarchical Spatio-Temporal Attention Networks

These #LangChain-powered apps include a Research Assistant that generates reports using web scraping and GPT-4o-mini, and Chat With Video, a #Streamlit app that #transcribes videos and enables content-based Q&A via #embeddings.

AI-powered Bilibili video Q&A agent — ask anything about any video in the comment section. 在 B 站评论区 @机器人，即可对任意视频提问。

[EMNLP 2025] D-CoDe: Scaling Image-Pretrained VLMs to Video via Dynamic Compression and Question Decomposition

These #LangChain-powered apps include a Research Assistant that generates reports using web scraping and GPT-4o-mini, and Chat With Video, a #Streamlit app that #transcribes videos and enables content-based Q&A via #embeddings.

DynFrame: Adaptive Reasoning-Driven Multimodal Framework with Dynamic Frame Augmentation for Complex Video Understanding

AI 科技口播与演示视频 Codex Skill：真实界面、官方素材、连续口播、分镜与逐镜头 QA | Evidence-first AI video workflow

pip install vidqa-cli — local, deterministic video QA CLI + MCP for tests, CI, and agents: 28 commands — frame timing, golden-frame diff, failure auto-locate, on-screen text index, color-aware run-to-run diff, Playwright/adb capture, local-VLM rubric review. One compact JSON + exit code per call. $0, fully on-device.

An open-source multimodal Video RAG framework for long-video understanding, action-level retrieval, ASR/OCR fusion and VLM reasoning.

These #LangChain-powered apps include a Research Assistant that generates reports using web scraping and GPT-4o-mini, and Chat With Video, a #Streamlit app that #transcribes videos and enables content-based Q&A via #embeddings.

[ACL BioNLP2026] LAMAR-2 at MedGenVidQA@BioNLP2026: Visual Answer Localization in Medical Videos via Multimodal LLM and Context-Augmented Prompting. Shared task website:

Local, zero-network CLI scoring AI short-drama clips for character-consistency drift and physics-plausibility flags