일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | |||
5 | 6 | 7 | 8 | 9 | 10 | 11 |
12 | 13 | 14 | 15 | 16 | 17 | 18 |
19 | 20 | 21 | 22 | 23 | 24 | 25 |
26 | 27 | 28 | 29 | 30 | 31 |
- deep learning
- r-cnn
- machine learning
- dfs
- real-time object detection
- C++
- opencv
- ubuntu
- 딥러닝
- 머신러닝
- hm3d
- MySQL
- NLP
- dynamic programming
- eecs 498
- 강화학습
- hm3dsem
- AlexNet
- Reinforcement Learning
- Python
- DP
- LSTM
- two-stage detector
- 그래프 이론
- YoLO
- CNN
- BFS
- 백준
- Mask Processing
- image processing
- Today
- Total
목록VQA (3)
JINWOOJUNG

Paperhttps://arxiv.org/abs/2303.11327 3D Concept Learning and Reasoning from Multi-View ImagesHumans are able to accurately reason in 3D by gathering multi-view observations of the surrounding world. Inspired by this insight, we introduce a new large-scale benchmark for 3D multi-view visual question answering (3DMV-VQA). This dataset is collected barxiv.org IntroductionVisual Reasoning은 시각적 장면에 ..

PapaerVQA: Visual Question Answering IntroductionImage&Video Captioning 분야에서의 Computer Vision&Natural Language Processing&Knowledge Representation을 결합한한 연구는 최근들어 많이 진행되고 있다. 기존 Image Captioning Task의 경우 Image의 완벽한 이해 없이도 간단한 Word Statistics만으로도 SOTA 성능을 보이는 것으로 보아 이미지의 깊은 이해 및 추론 성능이 불필요함을 의미한다. 즉, Image Captioning Task가 생각보다 "AI-complete"하지 않음을 시사한다.Knowledge Representation : 인간의 지식, 정보 등을 기계가 ..

Vision Language Model(VLM)VLM은 Computer Vision과 Natural Language Processing을 결합한 모델로써, 이미지나 비디오와 같은 시각적 데이터와 텍스트 입력을 동시에 이해하고 처리할 수 있다. VLM TasksVisual Question Answering(VQA) & Visual ReasoningVQA는 시각 질의응답을 의미하며, 이미지와 질문을 동시에 입력받아 정답을 출력하는 태스크이다. 아래 그림과 같이, 그림과 질문을 입력으로 받아 단답형 정답을 도출하는 형태이며, 이 태스크의 초기 벤치마크로는 [VQA: Visual Question Answering (Antol et al., 2015)] 논문이 있다. VQA v1(2015) Benchmark는 ..