https://github.com/zddhub/opensr
Open Sketch Recognize
https://github.com/zddhub/opensr
opengl opensse
Last synced: 5 months ago
JSON representation
Open Sketch Recognize
- Host: GitHub
- URL: https://github.com/zddhub/opensr
- Owner: zddhub
- License: other
- Created: 2014-06-12T10:38:34.000Z (almost 11 years ago)
- Default Branch: master
- Last Pushed: 2021-10-24T04:44:00.000Z (over 3 years ago)
- Last Synced: 2024-10-28T04:18:06.439Z (6 months ago)
- Topics: opengl, opensse
- Language: C++
- Homepage: https://sr.opensse.com
- Size: 30.4 MB
- Stars: 131
- Watchers: 8
- Forks: 18
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
Open Sketch Recognize
=====================
[Try via web](http://online.opensse.com).
This is another [OpenSSE demo](http://sr.opensse.com), supported by [OpenSSE](https://github.com/zddhub/opensse).
*Note:* Strictly speaking, this demo is based on **search** method, not **recognize** method now.
How to compile
==============
After installed [OpenSSE](https://github.com/zddhub/opensse) and [Qt 5.1+](https://www.qt.io), You can compile this like below:
```sh
cd gui/SketchRecognizeDemo
mkdir release && cd release
qmake -o Makefile ../SketchRecognizeDemo.pro
make
```
Demo - SketchRecognizeDemo
==========================
After compiled, you will have our SketchSearchDemo app under `bin` folder, open it like blow:
By the way, press 'C' in sketch pad, You will clear the sketch and draw next one.
See more recognize results, click [demo videos](http://sr.opensse.com).
Database
========
Sketch database comes from Mathias Eitz's paper [How Do Humans Sketch Objects? ACM Transactions on Graphics (Proc. SIGGRAPH 2012)](http://cybertron.cg.tu-berlin.de/eitz/projects/classifysketch/).
If you want to train your database, please refer to [OpenSSE Wiki](https://github.com/zddhub/opensse/wiki/How-to-train-data), you will get your own `sketches_filelist`、`sketches_index_file` and `vocabulary` file, and then update `config.json`.
Play with Tensorflow
====================
I have already extracted features from [Eitz's database](http://cybertron.cg.tu-berlin.de/eitz/projects/classifysketch/) under `data` folder. `sketches` includes 20000 images x 1000bins, the feature is based on `bag of words`, not pure feature.
If you install [Tensorflow](https://www.tensorflow.org) and you can play with:
```sh
python3 trytf.py
```
Thanks
======
- [Mathias Eitz](http://cybertron.cg.tu-berlin.de/eitz/)
License
=======
Code is under the [Apache License, Version 2.0](http://www.apache.org/licenses/LICENSE-2.0).
Donations
=========
I accept tips through [Alipay](assets/zddhub.png). Scan below qr code to donate:
Contact me
==========
If you have any question or idea, please [email to me](mailto:[email protected]).