[Detectron2] Densepose COCO 데이터셋

Point-based annotations:

• dp_x and dp_y: image coordinates of the annotated points along the horizontal and vertical axes respectively. The coordinates are defined with respect to the top-left corner of the annotated bounding box and are normalized assuming the bounding box size to be 256x256;
• dp_I: for each point specifies the index of the fine segmentation chart it belongs to;
• dp_U and dp_V: point coordinates on the corresponding chart. Each fine segmentation part has its own parametrization in terms of chart coordinates.

Segmentation annotations:

• dp_masks: RLE encoded dense masks (dict containing keys counts and size). The masks are typically of size 256x256, they define segmentation within the bounding box.

 

Chart-based Annotations

Chart-based DensePose COCO annotations are available for the instances of category person and correspond to the model shown in Figure 2. They include dp_x, dp_y, dp_I, dp_U and dp_V fields for annotated points (~100 points per annotated instance) and dp_masks field, which encodes coarse segmentation into 14 parts in the following order: Torso, Right Hand, Left Hand, Left Foot, Right Foot, Upper Leg Right, Upper Leg Left, Lower Leg Right, Lower Leg Left, Upper Arm Left, Upper Arm Right, Lower Arm Left, Lower Arm Right, Head.

Figure 2. Human body charts (fine segmentation) and the associated 14 body parts depicted with rounded rectangles (coarse segmentation).

 

The dataset splits used in the training schedules are train2014, valminusminival2014 and minival2014. train2014 and valminusminival2014 are used for training, and minival2014 is used for validation. The table with annotation download links, which summarizes the number of annotated instances and images for each of the dataset splits is given below:

 

Name# inst# imagesfile sizedownload

densepose_train2014	39210	26437	526M	densepose_train2014.json
densepose_valminusminival2014	7297	5984	105M	densepose_valminusminival2014.json
densepose_minival2014	2243	1508	31M	densepose_minival2014.json

 

 

 

GitHub - facebookresearch/detectron2: Detectron2 is FAIR's next-generation platform for object detection, segmentation and other