AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Benchmarking Omni-Vision Representation through the Lens of Visual Realms

Yuanhan Zhang, Zhenfei Yin, Jing Shao, Ziwei Liu

ECCV 2022 · Vision-language and multimodal › Multimodal LLMs · approach: VLM capability analysis · 21 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
use Swin-T instead of ResNet-50ImageNet-1K supervised · OmniBenchmarkavg top-1: 34.3 -> 46.4Improved
pre-train ResNet-50 on 69M GV-D dataResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 45.4Improved
use self-supervised DINO instead of supervised RN50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 38.9Improved
use MoCov2 instead of supervised RN50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 34.8No clear effect
use SwAV-Places pre-training instead of ImageNetResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 33.1Hurt
use masked image modeling (MAE) instead of supervised RN50ViT/ResNet-50 comparison · OmniBenchmarkavg top-1: 34.3 -> 30.6Hurt
use CutMix augmentation on ResNet-50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 31.1Hurt
use ReLabel on ResNet-50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 30.8Hurt
use MEAL-V2 distillation on ResNet-50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 36.6Improved
use PeCo RN50 instead of RN50ResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 36.4Improved
use ReCo RN101 instead of ResNet-101ResNet-101 · OmniBenchmarkavg top-1: 37.4 -> 38.7Improved
use CLIP image-text pre-trainingResNet-50 · OmniBenchmarkavg top-1: 34.3 -> 42.1Improved

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