HyperAI超神经

Object Counting On Fsc147

评估指标

MAE(test)

MAE(val)

RMSE(test)

RMSE(val)

评测结果

各个模型在此基准测试上的表现结果

					Paper Title
FamNet	22.08	23.75	99.54	69.07	Learning To Count Everything
Omnicount (Open vocabulary, multi-label, without training)	18.63	-	112	-	OmniCount: Multi-label Object Counting with Semantic-Geometric Priors
RCC	17.12	17.49	104.53	58.81	Learning to Count Anything: Reference-less Class-agnostic Counting with Weak Supervision
Counting-DETR	16.79	-	123.56	-	Few-shot Object Counting and Detection
CounTX (uses text descriptions instead of visual exemplars)	15.88	17.10	106.29	65.61	Open-world Text-specified Object Counting
LaoNet	15.78	17.11	97.15	56.81	Object Counting: You Only Need to Look at One
BMNet+	14.62	15.74	91.83	58.53	Represent, Compare, and Learn: A Similarity-Aware Framework for Class-Agnostic Counting
SAFECount	14.32	15.28	85.54	47.20	Few-shot Object Counting with Similarity-Aware Feature Enhancement
GCA-SUN	14.00	16.06	92.19	53.04	GCA-SUNet: A Gated Context-Aware Swin-UNet for Exemplar-Free Counting
SPDCN	13.51	14.59	96.80	49.97	Scale-Prior Deformable Convolution for Exemplar-Guided Class-Agnostic Counting
SemAug-SAFECount	12.74	12.59	89.90	44.95	Semantic Generative Augmentations for Few-Shot Counting
CounTR	11.95	13.13	91.23	49.83	CounTR: Transformer-based Generalised Visual Counting
SemAug-CounTR	11.32	12.31	77.50	41.65	Semantic Generative Augmentations for Few-Shot Counting
LOCA	10.79	10.24	56.97	32.56	A Low-Shot Object Counting Network With Iterative Prototype Adaptation
SSD	9.58	9.73	64.13	29.72	Learning Spatial Similarity Distribution for Few-shot Object Counting
CACViT	9.13	10.63	48.96	37.95	Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting
DAVE	8.66	8.91	32.36	28.08	DAVE -- A Detect-and-Verify Paradigm for Low-Shot Counting
GeCo	7.91	9.52	54.28	43.00	A Novel Unified Architecture for Low-Shot Counting by Detection and Segmentation
CountGD	5.74	7.1	24.09	26.08	CountGD: Multi-Modal Open-World Counting

0 of 19 row(s) selected.

HyperAI超神经

Object Counting On Fsc147

评估指标

MAE(test)

MAE(val)

RMSE(test)

RMSE(val)

评测结果

各个模型在此基准测试上的表现结果

					Paper Title
FamNet	22.08	23.75	99.54	69.07	Learning To Count Everything
Omnicount (Open vocabulary, multi-label, without training)	18.63	-	112	-	OmniCount: Multi-label Object Counting with Semantic-Geometric Priors
RCC	17.12	17.49	104.53	58.81	Learning to Count Anything: Reference-less Class-agnostic Counting with Weak Supervision
Counting-DETR	16.79	-	123.56	-	Few-shot Object Counting and Detection
CounTX (uses text descriptions instead of visual exemplars)	15.88	17.10	106.29	65.61	Open-world Text-specified Object Counting
LaoNet	15.78	17.11	97.15	56.81	Object Counting: You Only Need to Look at One
BMNet+	14.62	15.74	91.83	58.53	Represent, Compare, and Learn: A Similarity-Aware Framework for Class-Agnostic Counting
SAFECount	14.32	15.28	85.54	47.20	Few-shot Object Counting with Similarity-Aware Feature Enhancement
GCA-SUN	14.00	16.06	92.19	53.04	GCA-SUNet: A Gated Context-Aware Swin-UNet for Exemplar-Free Counting
SPDCN	13.51	14.59	96.80	49.97	Scale-Prior Deformable Convolution for Exemplar-Guided Class-Agnostic Counting
SemAug-SAFECount	12.74	12.59	89.90	44.95	Semantic Generative Augmentations for Few-Shot Counting
CounTR	11.95	13.13	91.23	49.83	CounTR: Transformer-based Generalised Visual Counting
SemAug-CounTR	11.32	12.31	77.50	41.65	Semantic Generative Augmentations for Few-Shot Counting
LOCA	10.79	10.24	56.97	32.56	A Low-Shot Object Counting Network With Iterative Prototype Adaptation
SSD	9.58	9.73	64.13	29.72	Learning Spatial Similarity Distribution for Few-shot Object Counting
CACViT	9.13	10.63	48.96	37.95	Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting
DAVE	8.66	8.91	32.36	28.08	DAVE -- A Detect-and-Verify Paradigm for Low-Shot Counting
GeCo	7.91	9.52	54.28	43.00	A Novel Unified Architecture for Low-Shot Counting by Detection and Segmentation
CountGD	5.74	7.1	24.09	26.08	CountGD: Multi-Modal Open-World Counting

0 of 19 row(s) selected.

Object Counting On Fsc147 | SOTA | HyperAI超神经