DAMUNT

Domain adaptive multi-ant tracking dataset (DAMUNT) is a dataset formulated to address the ant tracking problem with domain adaptation. Domain adaptation addresses the challenge of generating labelled data in rapidly evolving experiments involving social insects. The dataset consists of 57 video sequences that were recorded in a standard laboratory environment as part of an actual biological experiment. Serving as a domain adaptation dataset, it encompasses a source domain containing 33 training videos, along with a target domain comprising 18 training videos and 6 test videos. The primary objective of this challenge is to evaluate the performance of multi-object tracking on the target domain, where labels are not available for training.

Training Set

Sample Name FPS Resolution Length Tracks BoxesDensityDescriptionSourceRef.
CU10L1B4In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU30L1B6In_0251024x542500 (00:20)000.0A target domain sequence.link[1]
CU25L1B4In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU50L1B6Out_0251024x542500 (00:20)000.0A target domain sequence.link[1]
CU10L1B4Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU10L1B5Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU10L1B1In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU25L1B4Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU10L1B1Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU10L1B5In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU15L1B4Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU15L1B1Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU50L1B6In_0251024x542500 (00:20)000.0A target domain sequence.link[1]
CU20L1B1In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU25L1B1In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU10L1B6Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU20L1B4In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU25L1B1Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
OU10B1L3In_0251024x542500 (00:20)1632236.4A source domain sequence.link[1]
OU50B2L3In_0251024x542720 (00:29)533338346.4A source domain sequence.link[1]
OU50B2L2In_0251024x542500 (00:20)522385947.7A source domain sequence.link[1]
OU10B3L2In_0251024x542720 (00:29)10718710.0A source domain sequence.link[1]
OU10B1L1In_0251024x542500 (00:20)2637547.5A source domain sequence.link[1]
OU50B3L3In_0251024x542500 (00:20)492310546.2A source domain sequence.link[1]
OU10B3L2Out_0251024x542720 (00:29)10720010.0A source domain sequence.link[1]
OU10B1L2In_0251024x542500 (00:20)1319523.9A source domain sequence.link[1]
OU10B3L3Out_0251024x542720 (00:29)10720010.0A source domain sequence.link[1]
OU10B1L2Out_0251024x542500 (00:20)10500010.0A source domain sequence.link[1]
OU50B1L2Out_0251024x542500 (00:20)472297946.0A source domain sequence.link[1]
OU10B1L1Out_0251024x542500 (00:20)10500010.0A source domain sequence.link[1]
OU10B3L1In_0251024x542720 (00:29)1171209.9A source domain sequence.link[1]
OU50B1L3In_0251024x542500 (00:20)921795635.9A source domain sequence.link[1]
OU10B2L3In_0251024x542720 (00:29)1171109.9A source domain sequence.link[1]
OU10B1L3Out_0251024x542500 (00:20)10500010.0A source domain sequence.link[1]
OU50B1L3Out_0251024x542500 (00:20)492450049.0A source domain sequence.link[1]
OU50B3L3Out_0251024x542500 (00:20)492450049.0A source domain sequence.link[1]
OU10B2L2Out_0251024x542720 (00:29)10720010.0A source domain sequence.link[1]
OU50B1L2In_0251024x542500 (00:20)721764835.3A source domain sequence.link[1]
OU10B2L1Out_0251024x542720 (00:29)10720010.0A source domain sequence.link[1]
OU10B2L2In_0251024x542720 (00:29)1162108.6A source domain sequence.link[1]
OU50B3L1In_0251024x542500 (00:20)502257245.1A source domain sequence.link[1]
OU50B2L1In_0251024x542720 (00:29)553263845.3A source domain sequence.link[1]
OU10B2L1In_0251024x542720 (00:29)1750977.1A source domain sequence.link[1]
OU50B1L1Out_0251024x542500 (00:20)492355347.1A source domain sequence.link[1]
OU10B3L3In_0251024x542720 (00:29)1071109.9A source domain sequence.link[1]
OU50B2L2Out_0251024x542720 (00:29)503600050.0A source domain sequence.link[1]
OU50B3L2In_0251024x542500 (00:20)522299246.0A source domain sequence.link[1]
OU10B2L3Out_0251024x542720 (00:29)964809.0A source domain sequence.link[1]
OU50B3L2Out_0251024x542500 (00:20)492450049.0A source domain sequence.link[1]
OU10B3L1Out_0251024x542720 (00:29)10720010.0A source domain sequence.link[1]
OU50B1L1In_0251024x542500 (00:20)691497229.9A source domain sequence.link[1]
Total 32100 frm.
(1290 s.)
1051 471400 14.7

Test Set

Sample Name FPS Resolution Length Tracks BoxesDensityDescriptionSourceRef.
CU10L1B6In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU20L1B4Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU15L1B1In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU30L1B6Out_0251024x542500 (00:20)000.0A target domain sequence.link[1]
CU20L1B1Out_0251024x542720 (00:29)000.0A target domain sequence.link[1]
CU15L1B4In_0251024x542720 (00:29)000.0A target domain sequence.link[1]
Total 4100 frm.
(165 s.)
0 0 0.0


Download

Get train data (4.0 GB)
Get eval data (861,4 MB)
Development Kit

References:


[1] Abeysinghe, C., Reid, C., Rezatofighi, H. & Meyer, B. Tracking Different Ant Species: An Unsupervised Domain Adaptation Framework and a Dataset for Multi-object Tracking. IJCAI, 2023.