Code and papers for COMP2560 (Studies in Advanced Computing R&D) at ANU. This semester, my project is to implement a system for human pose estimation from videos. The aim of the system is to take as input as video of a person (or people) performing some interesting actions, then output an annotation for each video frame which shows the location of some key points on their body in that frame. For instance, you might define your key points to be the {left, right} {wrist, elbow, shoulder}, plus additional keypoints at the base of the neck and in the middle of the face. Knowing the locations of these body parts can make it easier to perform high-level tasks like gesture recognition.
Contents of this repository:
-
At the moment,
project/
houses code related to body part detection in single frames. This is implemented using the Caffe deep learning framework, and follows the approach of Chen & Yuille in Articulated Pose Estimation by a Graphical Model with Image Dependent Pairwise Relations.Initially, I attempted to re-implement Chen & Yuille's method in Python, but this proved to be too much work for a single-term project, so I instead opted to extend their Matlab code, along with that of a few other researchers. The result is in
project/matlab/
. -
thirdparty/
is for papers, code and data sets from other researchers. Some of those data sets are large, so I've usedgit-annex
to store them.
Copyright is complicated. Stuff in thirdparty/
was written by other
researchers, and licenses vary. The same goes for things in project/matlab
,
which was largely written by other researchers (Chen, Yuille, Yang, Ramanan,
Cherian, etc.) and adapted by me, so licensing varies there, too. Everything
else (including the Python code in project/
and the stuff in report/
) is
Apache v2 licensed:
Copyright 2015 Sam Toyer
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.