Skip to content

Latest commit

 

History

History
266 lines (225 loc) · 9.26 KB

README.md

File metadata and controls

266 lines (225 loc) · 9.26 KB
Raw

MP-100 is built upon the 2D pose datasets, including COCO, 300W, AFLW, OneHand10K, DeepFashion2, AP-10K, MacaquePose, Vinegar Fly, Desert Locust, CUB-200, CarFusion, AnimalWeb, Keypoint-5. In order to use MP-100, please download images from the original datasets first, then reorganize the data and use our provided annotation files for training and testing. After preparing images and annotations, the project should look like this:

Pose-for-Everything
├── assets
├── configs
├── mp100
├── pomnet
├── tools
`── data
    │── mp100
        │-- annotations
        │   │-- mp100_split1_train.json
        │   |-- mp100_split1_val.json
        │   |-- mp100_split1_test-dev-2017.json
        │   │-- ...
        │-- human_face
        │-- human_hand
        │-- sling_dress
        │-- human_body
        │   │-- 000000000009.jpg
        │   │-- 000000000025.jpg
        │   │-- 000000000030.jpg
        │   │-- ...
        │-- antelope_body
        │-- ...

MP-100 includes 100 categories and the images of different categories are contained in different folders individually. Specifically,

  • human_body is collected from COCO.

    Here is an example that soft link can be established from the downloaded images to propare the data for each category.

    ln -s ${COCO_PATH} data/mp100/human_body

  • human_face is collected from 300W.

  • amur_tiger_body is collected from AFLW.

  • human_hand is collected from OneHand10K.

  • 13 categories including long_sleeved_dress, long_sleeved_outwear, long_sleeved_shirt, shorts, short_sleeved_dress, short_sleeved_outwear, short_sleeved_shirt, skirt, sling, sling_dress, trousers, vest, and vest_dress are collected from DeepFashion2.

    For simplicity, all the categories can be linked to the complete downloaded dataset.

    ln -s ${DEEPFASHION_DATA} data/mp100/long_sleeved_dress
ln -s ${DEEPFASHION_DATA} data/mp100/long_sleeved_outwear
ln -s ${DEEPFASHION_DATA} data/mp100/long_sleeved_shirt
...

  • 34 categories including antelope_body, beaver_body, bison_body, bobcat_body, cat_body, cheetah_body, cow_body, deer_body, dog_body, elephant_body, fox_body, giraffe_body, gorilla_body, hamster_body, hippo_body, horse_body, leopard_body, lion_body, otter_body, panda_body, panther_body, pig_body, polar_bear_body, rabbit_body, raccoon_body, rat_body, rhino_body, sheep_body, skunk_body, spider_monkey_body, squirrel_body, weasel_body, wolf_body, and zebra_body are collected from AP-10K.

  • macaque_body is collected from MacaquePose.

  • fly_body is collected from Vinegar Fly.

  • locust_body is collected from Desert Locust.

  • 8 categories are collected from CUB-200. In detail,

    grebe_body is the combination of 050.Eared_Grebe, 051.Horned_Grebe, 052.Pied_billed_Grebe, and 053.Western_Grebe in CUB-200.

    gull_body is the combination of 059.California_Gull, 060.Glaucous_winged_Gull, 061.Heermann_Gull, 062.Herring_Gull, 063.Ivory_Gull, 064.Ring_billed_Gull, 065.Slaty_backed_Gull, and 066.Western_Gull in CUB-200.

    kingfisher_body is the combination of 079.Belted_Kingfisher, 080.Green_Kingfisher, 081.Pied_Kingfisher, 082.Ringed_Kingfisher, and 083.White_breasted_Kingfisher in CUB-200.

    sparrow_body is the combination of 113.Baird_Sparrow, 114.Black_throated_Sparrow, 115.Brewer_Sparrow, 116.Chipping_Sparrow, 117.Clay_colored_Sparrow, 118.House_Sparrow, 119.Field_Sparrow, 120.Fox_Sparrow, 121.Grasshopper_Sparrow, 122.Harris_Sparrow, 123.Henslow_Sparrow, 124.Le_Conte_Sparrow, 125.Lincoln_Sparrow, 126.Nelson_Sharp_tailed_Sparrow, 127.Savannah_Sparrow, 128.Seaside_Sparrow, 129.Song_Sparrow, 130.Tree_Sparrow, 131.Vesper_Sparrow, 132.White_crowned_Sparrow, and 133.White_throated_Sparrow in CUB-200.

    tern_body is the combination of 141.Artic_Tern, 142.Black_Tern, 143.Caspian_Tern, 144.Common_Tern, 145.Elegant_Tern, 146.Forsters_Tern, and 147.Least_Tern in CUB-200.

    warbler_body is the combination of 158.Bay_breasted_Warbler, 159.Black_and_white_Warbler, 160.Black_throated_Blue_Warbler, 161.Blue_winged_Warbler, 162.Canada_Warbler, 163.Cape_May_Warbler, 164.Cerulean_Warbler, 165.Chestnut_sided_Warbler, 166.Golden_winged_Warbler, 167.Hooded_Warbler, 168.Kentucky_Warbler, 169.Magnolia_Warbler, 170.Mourning_Warbler, 171.Myrtle_Warbler, 172.Nashville_Warbler, 173.Orange_crowned_Warbler, 174.Palm_Warbler, 175.Pine_Warbler, 176.Prairie_Warbler, 177.Prothonotary_Warbler, 178.Swainson_Warbler, 179.Tennessee_Warbler, 180.Wilson_Warbler, 181.Worm_eating_Warbler, and 182.Yellow_Warbler in CUB-200.

    woodpecker_body is the combination of 187.American_Three_toed_Woodpecker, 188.Pileated_Woodpecker, 189.Red_bellied_Woodpecker, 190.Red_cockaded_Woodpecker, 191.Red_headed_Woodpecker, and 192.Downy_Woodpecker in CUB-200.

    wren_body is the combination of 193.Bewick_Wren, 194.Cactus_Wren, 195.Carolina_Wren, 196.House_Wren, 197.Marsh_Wren, 198.Rock_Wren, and 199.Winter_Wren in CUB-200.

    As the images of the category come from multiple sources, we can copy or move all the needed images to the new folder. For example,

    mkdir grebe_body

# copy images to the new folder
cp ${CUB-200_ROOT}/*_Grebe/* data/mp100/grebe_body
or
# move images to the new folder
mv ${CUB-200_ROOT}/*_Grebe/* data/mp100/grebe_body

  • 3 categories including bus, car, and suv are collected from CarFusion. We clean the data and select the samples manually. Also, we rename the images to image_id.jpg using rename_carfusion_image.py. image_id is the ID of each image in COCO format obtained by the official tools.

    First, we can use the code provided by official tools to convert the annotations to COCO format. Then, we run rename_carfusion_image.py to rename the images.

    python mp100/rename_carfusion_image.py --ann_file ${COCO_FORMAT_ANNOTATION} \
  --img_src ${CARFUSION_DATA} --write_dir data/mp100

  • 30 categories including alpaca_face, arcticwolf_face, bighornsheep_face, blackbuck_face, bonobo_face, californiansealion_face, camel_face, capebuffalo_face, capybara_face, chipmunk_face, commonwarthog_face, dassie_face, fallowdeer_face, fennecfox_face, ferret_face, gentoopenguin_face, gerbil_face, germanshepherddog_face, gibbons_face, goldenretriever_face, greyseal_face, grizzlybear_face, guanaco_face, klipspringer_face, olivebaboon_face, onager_face, pademelon_face, proboscismonkey_face, przewalskihorse_face, and quokka_face are collected from AnimalWeb.

  • 5 categories including bed, chair, sofa, swivelchair, and table are collected from Keypoint-5.

Citation

If you find this dataset useful in your research, please consider cite:

@article{xu2022pose,
  title={Pose for Everything: Towards Category-Agnostic Pose Estimation},
  author={Xu, Lumin and Jin, Sheng and Zeng, Wang and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping and Wang, Xiaogang},
  booktitle={European Conference on Computer Vision (ECCV)},
  year={2022},
  month={October}
}