mikewang/padv2
收藏资源简介：
--- pretty_name: 'Padv2 Dataset - Part1' language: - en --- # Dataset Card for Padv2 Part1 ## Dataset Description **Official Repo:** https://github.com/lhc1224/OSAD_Net#-dataset-; **IMPORTANT Notes**: - This Huggingface dataset loads the Part1 of the Padv2 dataset, i.e., the PADv2_part1.zip; The file can also be downloaded from: https://uofi.box.com/s/1atjh3d2p82qyxm3gp11514006va0llq - Each instance in the loaded HF dataset contains the following fields: - `image_uid`: unique id to a dataset instanec - `image_path`: path to the raw rgb image - `depth_path`: path to the depth annotation of the image - `mask_path`: path to the object mask of the image - `affordance_type`: affordance type of the object in the image - `original_divisions`: there are three versions of divisions on the affordance types in the original dataset, this field stores the split ("train" or "test") of this instance in the three different divisions ("divide_1", "divide_2", "divide_3") **Paper Citation:** ``` @inproceedings{Oneluo, title={One-Shot Affordance Detection}, author={Hongchen Luo and Wei Zhai and Jing Zhang and Yang Cao and Dacheng Tao}, booktitle={IJCAI}, year={2021} } ``` ``` @article{luo2021one, title={One-Shot Object Affordance Detection in the Wild}, author={Zhai, Wei and Luo, Hongchen and Zhang, Jing and Cao, Yang and Tao, Dacheng}, journal={arXiv preprint arXiv:2108.03658}, year={2021} } ``` ## Dataset Summary With complex scenes and rich annotations, the PADv2 dataset can be used as a test bed to benchmark affordance detection methods and may also facilitate downstream vision tasks, such as scene understanding, action recognition, and robot manipulation. It contains 30k diverse images covering 39 affordance categories as well as 103 object categories from different scenes.
--- 数据集展示名: 'Padv2数据集 - 第一部分' 语言: - 英语 --- # Padv2数据集 - 第一部分 数据集卡片 ## 数据集描述 **官方仓库:** https://github.com/lhc1224/OSAD_Net#-dataset-; **重要说明:** - 本Hugging Face数据集加载Padv2数据集的第一部分，即PADv2_part1.zip；该文件也可通过以下链接下载：https://uofi.box.com/s/1atjh3d2p82qyxm3gp11514006va0llq - 加载后的HF数据集每条实例包含以下字段： - `image_uid`: 数据集实例的唯一标识符 - `image_path`: 原始RGB图像的存储路径 - `depth_path`: 该图像深度标注的存储路径 - `mask_path`: 该图像目标掩码的存储路径 - `affordance_type`: 图像中目标的可供性（affordance）类型 - `original_divisions`: 原始数据集针对可供性类型设有三种划分方案，本字段存储该实例在三种划分（"divide_1"、"divide_2"、"divide_3"）下的拆分集（"train"训练集或"test"测试集） **论文引用:** @inproceedings{Oneluo, title={单样本可供性检测}, author={Hongchen Luo and Wei Zhai and Jing Zhang and Yang Cao and Dacheng Tao}, booktitle={IJCAI}, year={2021} } @article{luo2021one, title={野外场景下的单样本目标可供性检测}, author={Zhai, Wei and Luo, Hongchen and Zhang, Jing and Cao, Yang and Tao, Dacheng}, journal={arXiv preprint arXiv:2108.03658}, year={2021} } ## 数据集概述 凭借复杂多元的场景与丰富完备的标注，PADv2数据集可作为可供性检测方法的基准测试平台，同时也可助力下游视觉任务，例如场景理解、动作识别以及机器人操控。 该数据集包含3万张多样化图像，涵盖39个可供性类别与103个来自不同场景的目标类别。
数据集卡片 for Padv2 Part1
数据集描述
官方仓库: https://github.com/lhc1224/OSAD_Net#-dataset-;
重要说明:
- 此Huggingface数据集加载Padv2数据集的Part1部分，即PADv2_part1.zip；该文件也可以从以下链接下载：https://uofi.box.com/s/1atjh3d2p82qyxm3gp11514006va0llq
- 加载的HF数据集中的每个实例包含以下字段：
image_uid: 数据集实例的唯一IDimage_path: 原始RGB图像的路径depth_path: 图像深度注释的路径mask_path: 图像对象掩码的路径affordance_type: 图像中对象的功能类型original_divisions: 原始数据集中有三种不同的功能类型划分，此字段存储该实例在三种不同划分（"divide_1", "divide_2", "divide_3"）中的划分（"train" 或 "test"）
论文引用:
@inproceedings{Oneluo, title={One-Shot Affordance Detection}, author={Hongchen Luo and Wei Zhai and Jing Zhang and Yang Cao and Dacheng Tao}, booktitle={IJCAI}, year={2021} }
@article{luo2021one, title={One-Shot Object Affordance Detection in the Wild}, author={Zhai, Wei and Luo, Hongchen and Zhang, Jing and Cao, Yang and Tao, Dacheng}, journal={arXiv preprint arXiv:2108.03658}, year={2021} }
数据集概述
PADv2数据集包含复杂的场景和丰富的注释，可作为基准测试功能检测方法的试验台，并可能促进下游视觉任务，如场景理解、动作识别和机器人操作。
它包含30k张多样化的图像，涵盖39种功能类别以及来自不同场景的103种对象类别。
