Obstacle-dataset OD

Dataset for fifteen types of obstacle detection, Because the data is too large to put in Github, please click the link to get it from Googledriver.Baidu CODE:0521

Classes ["stop_sign","person","bicycle","bus","truck","car","motorbike","reflective_cone","ashcan","warning_column","spherical_roadblock","pole","dog","tricycle","fire_hydrant"] Need to underline the category name

This obstacle dataset follows the format of the VOC dataset, if you are not clear about the VOC data format, please click THERE to view.

The dataset contains xml files in VOC format and .txt files for yolo training,as follows.

VOC format.

-JPEGImages
-Annotations
-ImageSets
　--Main
　　---train.txt
　　---test.txt
　　---val.txt
According to pictures and labels,as follows.
　　img-train for training Contains 5066 images　　ann-train
　　img-test for test Contains 1583 images　　　　　ann-test
　　img-val for validation Contains 1266 images　　　ann-val

yolo format．

lables　　　　#all picture labels<br>
label-train　　#labels for training<br>
label-test　　　#labels for testing<br>
label-val　　　　#labels for validation<br>

This dataset contains images from the VOC dataset, the COCO dataset, and the TT100K dataset,It also contains some pictures collected by the author's team in the field.

VOC

Vicente S, Carreira J, Agapito L, Batista J. Reconstructing PASCAL VOC. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. 2014:41-48.

COCO

Lin TY, Maire M, Belongie S, et al. Microsoft COCO: Common objects in context. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics). 2014;8693 LNCS(PART 5):740-755.

TT100K

Zhu Z, Liang D, Zhang S, Huang X, Li B, Hu S. Traffic-Sign Detection and Classification in the Wild. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. 2016;2016-December:2110-2118.

if you have any question,You can contact me according to the following email.
E-mail:[email protected]

The author continues to update this dataset.
If you feel helpful to you, please light up the stars.