Poselets: Body part detectors trained using 3D human pose annotations

12
Poselets Poselets capture part of the pose from a given viewpoint [Bourdev & Malik, ICCV09]

Transcript of Poselets: Body part detectors trained using 3D human pose annotations

Page 1: Poselets: Body part detectors trained using 3D human pose annotations

Poselets

Poselets capture part of the pose from a given viewpoint [Bourdev & Malik,

ICCV09]

Page 2: Poselets: Body part detectors trained using 3D human pose annotations

Poselets

Examples may differ visually but have common semantics [Bourdev & Malik,

ICCV09]

Page 3: Poselets: Body part detectors trained using 3D human pose annotations

Poselets

But how are we going to create training examples of poselets?

Page 4: Poselets: Body part detectors trained using 3D human pose annotations

How do we train a poselet for a given pose configuration?

Page 5: Poselets: Body part detectors trained using 3D human pose annotations

Finding correspondences at training time

Given part of a human pose

How do we find a similar pose configuration in the training set?

Page 6: Poselets: Body part detectors trained using 3D human pose annotations

We use keypoints to annotate the joints, eyes, nose, etc. of people

Left Hip

Left Shoulder

Finding correspondences at training time

Page 7: Poselets: Body part detectors trained using 3D human pose annotations

Residual Error

Finding correspondences at training time

Page 8: Poselets: Body part detectors trained using 3D human pose annotations

Training poselet classifiers

Residual Error:

0.15

0.20

0.10

0.35

0.15

0.85

1. Given a seed patch2. Find the closest patch for every

other person3. Sort them by residual error4. Threshold them

Page 9: Poselets: Body part detectors trained using 3D human pose annotations

Training poselet classifiers

1. Given a seed patch2. Find the closest patch for every

other person3. Sort them by residual error4. Threshold them5. Use them as positive training

examples to train a linear SVM with HOG features

Page 10: Poselets: Body part detectors trained using 3D human pose annotations

Which poselets should we train?

• Choose thousands of random windows, generate poselet candidates, train linear SVMs

• Select a small set of poselets that are:• Individually effective• Complementary

Page 11: Poselets: Body part detectors trained using 3D human pose annotations

Selecting a small set of complementary

poselets

Page 12: Poselets: Body part detectors trained using 3D human pose annotations

Poselets website

http://eecs.berkeley.edu/~lbourdev/poselets

The set of published poselet papers H3D data set + Matlab tools Java3D annotation tool + video

tutorial Matlab code to detect people using

poselets Our latest trained poselets