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 |
- cvpr 2024
- Stable Diffusion
- Segment Anything 설명
- iclr spotlight
- contrastive learning
- 논문 리뷰
- Prompt란
- deep learning 논문 리뷰
- iclr 논문 리뷰
- Segment Anything 리뷰
- 자기지도학습
- 논문리뷰
- Data-centric AI
- Computer Vision 논문 리뷰
- active learning
- VLM
- cvpr 논문 리뷰
- CVPR
- Computer Vision
- ssl
- Data-centric
- iclr 2024
- Multi-modal
- deep learning
- Self-supervised learning
- Meta AI
- Segment Anything
- Prompt Tuning
- ICLR
- ai 최신 논문
- Today
- Total
목록논문리뷰 (5)
Study With Inha

CVPR 2020, Hyperbolic Image Embeddings 논문 링크: https://arxiv.org/abs/1904.02239(https://arxiv.org/abs/1904.02239) Hyperbolic Image Embeddings Computer vision tasks such as image classification, image retrieval and few-shot learning are currently dominated by Euclidean and spherical embeddings, so that the final decisions about class belongings or the degree of similarity are made using linear hy ..

Meta AI (현 메타 에이아이, 구 페이스북), Facebook Research team (FAIR) IMAGEBIND: One Embedding Space To Bind Them All 논문 링크: https://arxiv.org/pdf/2305.05665.pdf 1. Introduction 최근 Segment Anything Model (SAM) 이라는 것을 발표한 Meta AI에서 또 다른 논문을 발표했다. 2023년 5월 9일 아카이브에 올라온 최신 논문인데, 신선한 아이디어를 제시하고 있어 리뷰할 논문으로 선정했다. 이 논문에서 제시하는 'ImageBind'는 여러가지 모달리티들의 embedding을 하나의 공통 space에 정렬함으로써 긴밀한 관계를 형성하고, 이를 통해 다양한 multi-..

MIT CSAIL, T. Wang et al, Dataset Distillation, 2018 논문 링크: https://arxiv.org/pdf/1811.10959.pdf 1. Introduction 일반적으로 딥러닝에서는 대용량의 데이터셋으로 큰 모델을 학습시키는 것이 좋은 성능을 내고 있다. 하지만 그 경우 많은 메모리, 노동력, 시간 등등의 자원이 필요하므로 효율성 측면에서 좋은 학습 방법이라고 말하기는 어렵다. 그래서 많은 연구진들이 적은 자원으로 최대한의 효율을 낼 수 있는 방법론들에 대한 연구를 진행했고, 본 논문에서 소개하는 'Dataset Distillation (데이터셋 증류)' 도 그 중 하나라고 볼 수 있다. 데이터셋 증류란 대규모 데이터셋을 대표되는 몇 장의 합성 이미지로 압축한 ..

Meta AI, Segment Anything, Alexander Kirillov et al.논문 링크: https://ai.facebook.com/research/publications/segment-anything/ (SAM 후속 논문 리뷰 링크) Meta AI, SAM 2: Segment Anything in Images and Videos 논문 리뷰 및 SAM2 설명 [Paper Review] Meta AI, SAM 2: Segment Anything in Images and Videos 논문 리뷰 및 SAM2 설명Meta FAIR SAM 2: Segment Anything in Images and Videos SAM2 설명 및 논문 리뷰Paper: SAM2 Paper LinkDemo: https..

Core-set: Active Learning for Convolutional Neural Networks 논문 링크: https://arxiv.org/abs/1708.00489 Active Learning for Convolutional Neural Networks: A Core-Set Approach Convolutional neural networks (CNNs) have been successfully applied to many recognition and learning tasks using a universal recipe; training a deep model on a very large dataset of supervised examples. However, this approach i..