Cross-domain Few-annotation Industrial Dataset
- Citation Author(s):
-
Yuhang HuangShilong ZouXinwang LiuKai Xu
- Submitted by:
- Shilong Zou
- Last updated:
- DOI:
- 10.21227/8jmr-gc98
- Data Format:
114 views
- Categories:
- Keywords:
Abstract
This dataset, mentioned in paper "MS2A: Memory Storage-to-Adaptation for Cross-domain Few-annotation Object Detection" and prepared for Cross-domain Few-annotation Object Detection task, consists of two cross-domain scenarios: Indus-S to Indus-T1 and Indus-S to Indus-T2. In detail, Indus-S consists of 4614 images for training and 1153 images for validation; Indus-T1 and Indus-T2 have 269 and 432 images for validation respectively. For the training data of Indus-T1 and Indus-T2, we introduce three different settings: 10-anno, 30-anno and 50-anno. This dataset was collected from different factories with different domains and labeled using LabelMe. The objects were annotated as the part class.
Dataset Link:
Baidu Netdisk link: https://pan.baidu.com/s/1QIVEVO5n1RYEGndHPe6aRg?pwd=cfod
Instructions:
The main directory of the dataset includes three different scenarios: Indus-S, Indus-T1, and Indus-T2. Each scenario file contains four files: annotations, train2017, val2017 and test2017. Specifically, the folder train2017, and val2017 are used for training and validation respectively, and the corresponding labels in COCO format are stored in the annotations folder.
z