mitmul / deeppose

DeepPose implementation in Chainer
http://static.googleusercontent.com/media/research.google.com/ja//pubs/archive/42237.pdf
GNU General Public License v2.0
408 stars 129 forks source link

Label data is used to Predict? #10

Open pradeepj247 opened 8 years ago

pradeepj247 commented 8 years ago

In the predict_flic.py, when I run it in test mode, it seems to load the label data for a particular file, and then does the prediction. (that's my understanding)

how will this work if I give a new image which is not there in the existing labelled data set. is there a simple example of how I can just take 1 image and get a prediction of joints.

mitmul commented 8 years ago

Thank you for using my codes. Basically the current prediction script is designed for evaluation and visualization, so it needs to load label data. It calculates mean Euclidean distance between correct joints and predicted joints. But I will put a simple script to predict from a single image that is given by command args.

kshalini commented 8 years ago

@mitmul thanks for the update and yes the other issue #11 is related to this.

just wondering if you got a chance to come up with this script to predict from any sample test image (using command line args). pls update here when you do. thanks!

nlgranger commented 8 years ago

Hi @mitmul I would be intersted in this too.

kristellmarisse commented 8 years ago

@mitmul : Any updates on this would be helpful.

ghost commented 8 years ago

@mitmul Any updates?