|
1 | 1 | # Release Notes |
2 | 2 |
|
| 3 | +## 0.2.0 (2016/01/19) |
| 4 | ++ See [this blog post](http://bamos.github.io/2016/01/19/openface-0.2.0/) |
| 5 | + for an overview, |
| 6 | + [the GitHub Milestone](https://github.com/cmusatyalab/openface/milestones/v0.2.0) |
| 7 | + for a high-level issue summary. |
| 8 | ++ Training improvements from resulting in an accuracy increase from **76.1% to 92.9%**, |
| 9 | + which are from Bartosz Ludwiczuk's ideas and implementations in |
| 10 | + [this mailing list thread](https://groups.google.com/forum/#!topic/cmu-openface/dcPh883T1rk). |
| 11 | + These improvements also reduce the training time from a week to a day. |
| 12 | ++ Nearly halved execution time thanks to [Hervé Bredin's](http://herve.niderb.fr/) |
| 13 | + suggestions and sample code for image alignment in |
| 14 | + [Issue 50](https://github.com/cmusatyalab/openface/issues/50). |
| 15 | ++ Hosted |
| 16 | + [Python API Documentation](http://openface-api.readthedocs.org/en/latest/index.html). |
| 17 | ++ [Docker automated build](https://hub.docker.com/r/bamos/openface) online. |
| 18 | ++ Initial automatic tests written in [tests](https://github.com/cmusatyalab/openface/tree/0.2.0/tests). |
| 19 | ++ [Tests successfully passing](https://travis-ci.org/cmusatyalab/openface/branches) |
| 20 | + in the Docker automated build in Travis. |
| 21 | ++ Add `util/profile-pipeline.py` |
| 22 | + [util/profile-pipeline.py](https://github.com/cmusatyalab/openface/tree/0.2.0/util/profile-pipeline.py) |
| 23 | + to profile the overall execution time on a single image. |
| 24 | + |
3 | 25 | ## 0.1.1 (2015/10/15) |
4 | 26 | + Fix debug mode of NaiveDlib alignment. |
5 | | -+ Add util/prune-dataset.py for dataset processing. |
| 27 | ++ Add |
| 28 | + [util/prune-dataset.py](https://github.com/cmusatyalab/openface/tree/0.1.1/util/prune-dataset.py) |
| 29 | + for dataset processing. |
6 | 30 | + Correct Docker dependencies. |
7 | 31 |
|
8 | 32 | ## 0.1.0 (2015/10/13) |
|
0 commit comments