Object Detection2019en

Large Vocabulary Instance Segmentation v1.0

1,203 object categories with federated, long-tail distribution across 164K COCO images. Tests real-world detection with rare and fine-grained categories.

Samples:164,000

Metrics:mask-ap, mask-ap50, mask-ap75, mask-ap-rare, mask-ap-common, mask-ap-frequent, box-ap

Paper / Website

Current State of the Art

DINO-X

IDEA Research

mask-ap

mask-ap Progress Over Time

Showing 5 breakthroughs from Dec 2021 to Nov 2024

Key Milestones

Dec 2021

Mask2Former (Swin-L)

Mask2Former with Swin-L backbone. 56.1 mask AP on LVIS v1.0 minival. Table 7, arxiv:2112.01527. CVPR 2022.

56.1

Source

Mar 2022

ViTDet-H (MAE)

ViTDet-H with MAE pretraining and Cascade Mask RCNN head. 59.5 mask AP on LVIS v1.0 minival. Table 3, arxiv:2203.16527. ECCV 2022.

59.5

+6.1%

Source

Nov 2022

InternImage-H

InternImage-H (1B params) with DINO detection head. 65.4 mask AP on LVIS v1.0 minival. Table 5, arxiv:2211.05778. CVPR 2023.

65.4

+9.9%

Source

Dec 2023

APE-Large

APE-Large with EVA-02 ViT-L backbone. 66.4 mask AP on LVIS v1.0 minival in closed-set evaluation. Table 2, arxiv:2312.02153. CVPR 2024.

66.4

+1.5%

Source

Nov 2024

DINO-XCurrent SOTA

DINO-X unified model. 67.0 mask AP on LVIS v1.0 minival — SOTA at time of release Nov 2024. Table 3, arxiv:2411.14347.

67.0

+0.9%

Source

Total Improvement

19.4%

Time Span

Breakthroughs

Current SOTA

67.0

Top Models Performance Comparison

Top 9 models ranked by mask-ap

Best Score

67.0

Top Model

DINO-X

Models Compared

Score Range

22.4

box-ap

#	Model	Score	Paper / Code	Date
1	DINO-XOpen Source IDEA Research	71.4	DINO-X: A Unified Vision Model for Open-World Object Detection and Understanding	Nov 2024
2	APE-LargeOpen Source Tsinghua / MEGVII	70.3	APE: Aligning and Prompting Everything All at Once for Universal Visual Perception	Dec 2023
3	Co-DINO (ViT-L)Open Source Sensetime / Sense-X	68	DETRs with Collaborative Hybrid Assignments Training	Nov 2022
4	ViTDet-H (MAE)Open Source Meta AI	64	Exploring Plain Vision Transformer Backbones for Object Detection	Mar 2022

mask-apPrimary

#	Model	Score	Paper / Code	Date
1	DINO-XOpen Source IDEA Research	67	DINO-X: A Unified Vision Model for Open-World Object Detection and Understanding	Nov 2024
2	APE-LargeOpen Source Tsinghua / MEGVII	66.4	APE: Aligning and Prompting Everything All at Once for Universal Visual Perception	Dec 2023
3	InternImage-HOpen Source Shanghai AI Lab	65.4	InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions	Nov 2022
4	EVA-02-LOpen Source BAAI / PKU	62.1	EVA-02: A Visual Representation Powerhouse for Dense Recognition Tasks	Mar 2023
5	ViTDet-H (MAE)Open Source Meta AI	59.5	Exploring Plain Vision Transformer Backbones for Object Detection	Mar 2022
6	Mask2Former (Swin-L)Open Source Meta AI / UIUC	56.1	Masked-attention Mask Transformer for Universal Image Segmentation	Dec 2021
7	ViTDet-HOpen Source Meta AI	53.4	arxiv	Mar 2026
8	EVA-02-L (LVIS)Open Source BAAI	50.7	arxiv	Mar 2026
9	Mask2Former (Swin-L) LVISOpen Source Meta AI	44.6	arxiv	Mar 2026

mask-ap-rare

#	Model	Score	Paper / Code	Date
1	APE-LargeOpen Source Tsinghua / MEGVII	65.4	APE: Aligning and Prompting Everything All at Once for Universal Visual Perception	Dec 2023
2	EVA-02-LOpen Source BAAI / PKU	61	EVA-02: A Visual Representation Powerhouse for Dense Recognition Tasks	Mar 2023
3	Mask2Former (Swin-L)Open Source Meta AI / UIUC	53.5	Masked-attention Mask Transformer for Universal Image Segmentation	Dec 2021