AI Paper Atlas

Research map / Detection and segmentation / CNN features

Could Giant Pre-trained Image Models Extract Universal Representations?

Yutong Lin, Ze Liu, Zheng Zhang, Han Hu, Nanning Zheng, Stephen Lin, Yue Cao

NeurIPS 2022 · Detection and segmentation › CNN features · approach: Convolution-transformer backbones · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Replace supervised ImageNet-1K pretraining with SimMIM masked image modeling (ablation)frozen Swin · COCOAP box: 42.4 -> 34.1Hurt
Pretrain SimMIM then fully finetune instead of freezing (ablation)SimMIM-1K · COCOAP box: 34.1 -> 52.9Improved
Scale supervised pretraining data from ImageNet-1K to ImageNet-22K (ablation)frozen Swin · Kinetics-400top-1 accuracy: 60.4 -> 70.3Improved
Replace supervised with contrastive EsViT pretraining on ImageNet-1K (ablation)frozen Swin · COCOAP box: 42.4 -> 42.0No clear effect
Replace FPN with BiFPN in the head (ablation)frozen SwinV2-B-22K · COCOAP box: 45.0 -> 51.9Improved
Add four residual blocks inside FPN (ablation)frozen SwinV2-B-22K · COCOAP box: 45.0 -> 49.5Improved
Replace the box/mask head with a cascade head (ablation)frozen SwinV2-B-22K · COCOAP box: 45.0 -> 49.0Improved
Enlarge pixel decoder from 1 to 6 blocks (ablation)frozen Mask2Former · ADE20KmIoU: 51.0 -> 52.6Improved
Enlarge transformer decoder from 4 to 10 blocks (ablation)frozen Mask2Former · ADE20KmIoU: 51.0 -> 51.9Improved
Replace linear head with four temporal Transformer blocks (ablation)frozen SwinV2-B-22K · Kinetics-400top-1 accuracy: 70.3 -> 74.7Improved
Use four global spatiotemporal blocks instead of temporal blocks (ablation)frozen SwinV2-B-22K · Kinetics-400top-1 accuracy: 74.7 -> 77.5Improved
Add 3 BiFPN layers to head under 1% COCO data (frozen vs full finetune) (ablation)frozen vs full ft. SwinV2-B-22K · COCO 1%AP box: 32.1 vs 25.9Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy