dddzg/up-detr
[TPAMI 2022 & CVPR2021 Oral] UP-DETR: Unsupervised Pre-training for Object Detection with Transformers
Introduces a random query patch detection pretext task for unsupervised transformer pre-training, eliminating annotation requirements during the initial phase while leveraging SwAV-initialized CNN backbones. Built on the DETR codebase with ResNet-50 backbone and transformer encoder-decoder, it achieves 43.1 AP on COCO after 300-epoch fine-tuning, outperforming supervised ImageNet pre-training with comparable training costs.
489 stars. No commits in the last 6 months.
Stars
489
Forks
72
Language
Python
License
Apache-2.0
Category
Last pushed
Jul 19, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/dddzg/up-detr"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related models
xiuqhou/Salience-DETR
[CVPR 2024] Official implementation of the paper "Salience DETR: Enhancing Detection Transformer...
hanouticelina/deformable-DETR
Implementation of the paper : Deformable DETR: Deformable Transformers for End-to-End Object...
xiuqhou/Relation-DETR
[ECCV2024 Oral] Official implementation of the paper "Relation DETR: Exploring Explicit Position...
akshitac8/OW-DETR
[CVPR 2022] Official Pytorch code for OW-DETR: Open-world Detection Transformer
BarCodeReader/SelfReformer
[TMM-2023] Official implementation of "Towards Complete and Detail-Preserved Salient Object...