Datasets:

karthik789338
/

OccDet-Calib

occlusion_level float64 0 0.8	num_predictions int64 551 7.19k	num_ground_truth int64 460 2k	tp int64 129 1.63k	fp int64 422 5.68k	fn int64 110 1.12k	precision float64 0.13 0.39	recall float64 0.26 0.86	f1 float64 0.2 0.51	model stringclasses 3 values	experiment stringclasses 3 values
0	7,167	2,000	1,625	5,542	375	0.226734	0.8125	0.354533	YOLOv8m	overlap_main
0.2	7,194	2,000	1,546	5,648	454	0.214901	0.773	0.336306	YOLOv8m	overlap_main
0.4	7,104	2,000	1,457	5,647	543	0.205096	0.7285	0.320079	YOLOv8m	overlap_main
0.6	6,975	2,000	1,297	5,678	703	0.18595	0.6485	0.289025	YOLOv8m	overlap_main
0.8	6,454	2,000	877	5,577	1,123	0.135885	0.4385	0.207476	YOLOv8m	overlap_main
0	3,397	1,000	829	2,568	171	0.244039	0.829	0.377075	FCOS-R50	overlap_main
0.2	3,431	1,000	776	2,655	224	0.226173	0.776	0.35026	FCOS-R50	overlap_main
0.4	3,461	1,000	733	2,728	267	0.211789	0.733	0.328626	FCOS-R50	overlap_main
0.6	3,334	1,000	642	2,692	358	0.192561	0.642	0.296262	FCOS-R50	overlap_main
0.8	3,100	1,000	462	2,638	538	0.149032	0.462	0.225366	FCOS-R50	overlap_main
0	3,980	1,000	857	3,123	143	0.215327	0.857	0.344177	Deformable-DETR-R50	overlap_main
0.2	4,001	1,000	833	3,168	167	0.208198	0.833	0.333133	Deformable-DETR-R50	overlap_main
0.4	3,982	1,000	789	3,193	211	0.198142	0.789	0.31674	Deformable-DETR-R50	overlap_main
0.6	3,942	1,000	682	3,260	318	0.173009	0.682	0.276002	Deformable-DETR-R50	overlap_main
0.8	3,676	1,000	468	3,208	532	0.127312	0.468	0.200171	Deformable-DETR-R50	overlap_main
0	1,294	500	390	904	110	0.301391	0.78	0.434783	YOLOv8m	distractor_control
0.2	1,319	490	377	942	113	0.285823	0.769388	0.416805	YOLOv8m	distractor_control
0.4	1,285	479	364	921	115	0.283268	0.759916	0.412698	YOLOv8m	distractor_control
0.6	1,266	472	358	908	114	0.28278	0.758475	0.411968	YOLOv8m	distractor_control
0.8	1,245	460	350	895	110	0.281124	0.76087	0.410557	YOLOv8m	distractor_control
0	1,294	500	390	904	110	0.301391	0.78	0.434783	YOLOv8m	truncation_control
0.2	963	500	374	589	126	0.38837	0.748	0.511278	YOLOv8m	truncation_control
0.4	876	500	346	530	154	0.394977	0.692	0.502907	YOLOv8m	truncation_control
0.6	739	500	262	477	238	0.354533	0.524	0.422922	YOLOv8m	truncation_control
0.8	551	500	129	422	371	0.23412	0.258	0.24548	YOLOv8m	truncation_control

OccDet-Calib

OccDet-Calib is a benchmark and analysis framework for studying detector calibration under partial visibility.

This repo contains

benchmark metadata CSVs
processed evaluation tables
paper figures and montage images
sample public assets
reproducibility-facing artifacts

This repo does not contain

raw COCO images
raw BDD100K images
checkpoints
virtual environments or cache folders

Main conditions

overlap occlusion
distractor control
truncation control
inpainting robustness subset
natural validation on BDD100K-style annotations

Main classes

person
bicycle
motorcycle
car
bus
truck
stop sign

Notes

The dataset viewer is configured with multiple subsets because the metadata and evaluation CSVs have different schemas.

Downloads last month: 25