PASCAL-Part Dataset

This dataset is a set of additional annotations for PASCAL VOC 2010. It goes beyond the original PASCAL object detection task by providing segmentation masks for each body part of the object. For categories that do not have a consistent set of parts (e.g., boat), we provide the silhouette annotation.

Below are some example segmentation annotations from the dataset.

Download

Training and Validation Set

The following download contains only the annotations for the training/validation set and some demo code.

trainval.tar.gz

  • MD5: 2fa0a19ee9b5e43b2bee520166111120

Original images for the dataset can be downloaded from the PASCAL VOC 2010 website.

Testing Set

Please download the images from the PASCAL VOC 2010 website.

back to top

Statistics

Since the dataset is an annotation of the PASCAL VOC 2010, it has the same statistics as those of the original dataset. Training and validation contains 10,103 images while testing contains 9,637 images.

Click on the panel below to see the list of annotated parts for each of the 20 PASCAL object classes.

Objects Parts
aeroplane
body
engine*
left wing
right wing
stern
tail
wheel*
bicycle
back wheel
chain wheel
front wheel
handlebar
headlight*
saddle
bird
beak
head
left eye
left foot
left leg
left wing
neck
right eye
right foot
right leg
right wing
tail
torso
boat
bottle
body
cap
bus
back license plate
back side
door*
front license plate
front side
headlight*
left mirror
left side
right mirror
right side
roof side
wheel*
window*
car
back license plate
back side
door*
front license plate
front side
headlight*
left mirror
left side
right mirror
right side
roof side
wheel*
window*
cat
head
left back leg
left back paw
left ear
left eye
left front leg
left front paw
neck
nose
right back leg
right back paw
right ear
right eye
right front leg
right front paw
tail
torso
chair
cow
head
left back lower leg
left back upper leg
left ear
left eye
left front lower leg
left front upper leg
left horn
muzzle
neck
right back lower leg
right back upper leg
right ear
right eye
right front lower leg
right front upper leg
right horn
tail
torso
diningtable
dog
head
left back leg
left back paw
left ear
left eye
left front leg
left front paw
muzzle
neck
nose
right back leg
right back paw
right ear
right eye
right front leg
right front paw
tail
torso
horse
head
left back hoof
left back lower leg
left back upper leg
left ear
left eye
left front hoof
left front lower leg
left front upper leg
muzzle
neck
right back hoof
right back lower leg
right back upper leg
right ear
right eye
right front hoof
right front lower leg
right front upper leg
tail
torso
motorbike
back wheel
front wheel
handlebar
headlight*
saddle
person
hair
head
left ear
left eye
left eyebrow
left foot
left hand
left lower arm
left lower leg
left upper arm
left upper leg
mouth
neck
nose
right ear
right eye
right eyebrow
right foot
right hand
right lower arm
right lower leg
right upper arm
right upper leg
torso
pottedplant
plant
pot
sheep
head
left back lower leg
left back upper leg
left ear
left eye
left front lower leg
left front upper leg
left horn
muzzle
neck
right back lower leg
right back upper leg
right ear
right eye
right front lower leg
right front upper leg
right horn
tail
torso
sofa
train
coach back side*
coach front side*
coach left side*
coach right side*
coach roof side*
coach*
head
head back side
head front side
head left tside
head right side
head roof side
headlight*
tvmonitor
screen

(* denotes a part can appear multiple times in one object instance)

close class list

back to top

Usage Considerations

We provide segmentation masks for detailed body parts. One can merge several parts to get appropriate object part granularity for different tasks. For instance, "eyes", "ears", "nose", etc. can be merged into a single "head" part.

back to top

Citation

back to top

Acknowledgements

We thank Viet Nguyen for coordinating and leading the efforts for cleaning up the annotations. We would like to acknowledge the support by grants ARO 62250-CS and N00014-12-1-0883.

back to top