SOTAVerified

Object Counting

The goal of Object Counting task is to count the number of object instances in a single image or video sequence. It has many real-world applications such as traffic flow monitoring, crowdedness estimation, and product counting.

Source: Learning to Count Objects with Few Exemplar Annotations

Papers

Showing 110 of 158 papers

TitleStatusHype
Car Object Counting and Position Estimation via Extension of the CLIP-EBC FrameworkCode0
OmniSpatial: Towards Comprehensive Spatial Reasoning Benchmark for Vision Language Models0
Improving Contrastive Learning for Referring Expression CountingCode0
Expanding Zero-Shot Object Counting with Rich Prompts0
InstructSAM: A Training-Free Framework for Instruction-Oriented Remote Sensing Object RecognitionCode2
Are Multimodal Large Language Models Ready for Omnidirectional Spatial Reasoning?0
VisionReasoner: Unified Visual Perception and Reasoning via Reinforcement LearningCode4
Learning What NOT to Count0
Marmot: Multi-Agent Reasoning for Multi-Object Self-Correcting in Improving Image-Text Alignment0
MATHGLANCE: Multimodal Large Language Models Do Not Know Where to Look in Mathematical Diagrams0
Show:102550
← PrevPage 1 of 16Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1FamNetMAE(test)22.08Unverified
2Omnicount (Open vocabulary, multi-label, without training)MAE(test)18.63Unverified
3RCCMAE(test)17.12Unverified
4Counting-DETRMAE(test)16.79Unverified
5CounTX (uses text descriptions instead of visual exemplars)MAE(test)15.88Unverified
6LaoNetMAE(test)15.78Unverified
7BMNet+MAE(test)14.62Unverified
8SAFECountMAE(test)14.32Unverified
9GCA-SUNMAE(test)14Unverified
10SPDCNMAE(test)13.51Unverified