【NLP2CV】 From NLP to Computer Vision

11 Apr 2022 in Artificial Intelligence

• From NLP to Computer Vision

1. Good Reference

1. (Updating) Awesome BERT & Transfer Learning in NLP
   • BERT related papers
   • BERT related articles
     • Attention
     • Transformer
     • GPT (Generative pre-training Transformer)
     • CLIP Contrastive Language-Image Pretraining
     • DALL-E 2 (Text-to-Image Revolution)
2. Transformer Attention Video
3. (Old) Awesome BERT
4. Transformer understanding by Me (1) attention
5. Transformer understanding by Me (2) Transformer
6. Awesome-CLIP: See Application contents (Detection)

2. Youtube Contents

1. Pixel-BERT: Aligning Image Pixels with Text by Deep Multi-Modal Transformers
2. BERT and GPT (한국어 설명)
3. BERT

3. Must Paper lists

Paper	Date	Institute, Citation	Note
Masked Autoencoders Are Scalable Vision Learners (MAE)	21.11	Kaiming He
BEiT: BERT Pre-Training of Image Transformers	21.06		MAE cited.
BERT: Pre-training of deep bidirectional transformers for language understanding	18.10		See youtube.
Improving language understanding by generative pretraining	18.11	OpenAI	GPT-1. MAE cited.
Language Models are Unsupervised Multitask Learners	19.09	OpenAI	GPT-2. MAE cited.
Language Models are Few-Shot Learners	20.05	OpenAI	GPT-3. MAE cited.
Contrastive Language-Image Pre-Training	21.01	OpenAI	CLIP
Zero-Shot Text-to-Image Generation	21.02		DALL-E
Hierarchical Text-Conditional Image Generation with CLIP Latents	22.04		DALL-E 2. To leverage CLIP representations for
Pathways Language Model (PaLM): Scaling to 540 Billion Parameters for Breakthrough Performance	22.04	Google Research	Survey on NLP
Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision	22.02	FAIR	SEER 10 parameter

4. MAE related papers

Paper	Date	Institute, Citation	Note
Context autoencoder for self-supervised representation learning
Tfill: Image completion via a transformer-based architecture
ibot: Image bert pre-training with online tokenizer
Benchmarking detection transfer learning with vision transformers
Corrupted image modeling for self-supervised visual pre-training
PeCo: Perceptual Codebook for BERT Pre-training of Vision Transformers
Bevt: Bert pretraining of video transformers
ViTAEv2: Vision Transformer Advanced by Exploring Inductive Bias for Image Recognition and Beyond
ReMixer: Object-aware Mixing Layer for Vision Transformers and Mixers
A Survey on Dropout Methods and Experimental Verification in Recommendation