jeonghyo.k
3946ff6a25
- prediction/image/ FastAPI 서버 Docker 환경 구성 - Dockerfile: PyTorch 2.1 + CUDA 12.1 기반 GPU 이미지 - docker-compose.yml: GPU 할당 + 데이터 볼륨 마운트 - requirements.txt: 서버 의존성 목록 - .env.example: 환경변수 템플릿 - DOCKER_USAGE.md: 빌드/실행/API 사용법 문서 - Dockerfile에 .dockerignore 제외 폴더 mkdir -p 추가 - .gitignore: prediction/image 결과물 및 모델 가중치(.pth) 제외 추가 - dbInsert_csv.py, dbInsert_shp.py 삭제 (미사용 DB 로직) - api.py: dbInsert import 및 주석 처리된 DB 호출 코드 제거 - aerialRouter.ts: req.params 타입 오류 수정
9.6 KiB
Vision Transformer
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Introduction
Abstract
While the Transformer architecture has become the de-facto standard for natural language processing tasks, its applications to computer vision remain limited. In vision, attention is either applied in conjunction with convolutional networks, or used to replace certain components of convolutional networks while keeping their overall structure in place. We show that this reliance on CNNs is not necessary and a pure transformer applied directly to sequences of image patches can perform very well on image classification tasks. When pre-trained on large amounts of data and transferred to multiple mid-sized or small image recognition benchmarks (ImageNet, CIFAR-100, VTAB, etc.), Vision Transformer (ViT) attains excellent results compared to state-of-the-art convolutional networks while requiring substantially fewer computational resources to train.
Citation
@article{dosoViTskiy2020,
title={An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale},
author={DosoViTskiy, Alexey and Beyer, Lucas and Kolesnikov, Alexander and Weissenborn, Dirk and Zhai, Xiaohua and Unterthiner, Thomas and Dehghani, Mostafa and Minderer, Matthias and Heigold, Georg and Gelly, Sylvain and Uszkoreit, Jakob and Houlsby, Neil},
journal={arXiv preprint arXiv:2010.11929},
year={2020}
}
Usage
To use other repositories' pre-trained models, it is necessary to convert keys.
We provide a script vit2mmseg.py in the tools directory to convert the key of models from timm to MMSegmentation style.
python tools/model_converters/vit2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
E.g.
python tools/model_converters/vit2mmseg.py https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_base_p16_224-80ecf9dd.pth pretrain/jx_vit_base_p16_224-80ecf9dd.pth
This script convert model from PRETRAIN_PATH and store the converted model in STORE_PATH.
Results and models
ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
|---|---|---|---|---|---|---|---|---|---|
| UPerNet | ViT-B + MLN | 512x512 | 80000 | 9.20 | 6.94 | 47.71 | 49.51 | config | model | log |
| UPerNet | ViT-B + MLN | 512x512 | 160000 | 9.20 | 7.58 | 46.75 | 48.46 | config | model | log |
| UPerNet | ViT-B + LN + MLN | 512x512 | 160000 | 9.21 | 6.82 | 47.73 | 49.95 | config | model | log |
| UPerNet | DeiT-S | 512x512 | 80000 | 4.68 | 29.85 | 42.96 | 43.79 | config | model | log |
| UPerNet | DeiT-S | 512x512 | 160000 | 4.68 | 29.19 | 42.87 | 43.79 | config | model | log |
| UPerNet | DeiT-S + MLN | 512x512 | 160000 | 5.69 | 11.18 | 43.82 | 45.07 | config | model | log |
| UPerNet | DeiT-S + LN + MLN | 512x512 | 160000 | 5.69 | 12.39 | 43.52 | 45.01 | config | model | log |
| UPerNet | DeiT-B | 512x512 | 80000 | 7.75 | 9.69 | 45.24 | 46.73 | config | model | log |
| UPerNet | DeiT-B | 512x512 | 160000 | 7.75 | 10.39 | 45.36 | 47.16 | config | model | log |
| UPerNet | DeiT-B + MLN | 512x512 | 160000 | 9.21 | 7.78 | 45.46 | 47.16 | config | model | log |
| UPerNet | DeiT-B + LN + MLN | 512x512 | 160000 | 9.21 | 7.75 | 45.37 | 47.23 | config | model | log |