license + IOU results

[ahundt]

Any chance you could give this a license? One good possible option would be the MIT license https://tldrlegal.com/license/mit-license, this is what Keras itself uses. In fact they might be interested in a pull request of this code to the official keras-contrib repository. https://github.com/farizrahman4u/keras-contrib

Also what kind of IOU results did you get with the models you define in this repository?

Thanks!

[aurora95]

Sorry I've been busy these days... I'll look into the license, but before I do anything I think I should first update this code to keras2.

In my experiments, AtrousFCN_Resnet50_16s got 65.75% IOU, and in #4 Fansiee got 66.1076%

[ahundt]

@aurora95 I've been able to run it on Keras-2 without changes! I only tried the Atrous Resnet version however.

Here is the license text by the way, it is pretty simple and lets people use it as they would like:

The MIT License (MIT)

Copyright (c) <year> <copyright holders>

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

I got this however (after loading pre-trained weights):

PASCAL VOC trained with pretrained imagenet weights
IOU:
[ 0.90873648  0.74772504  0.44416247  0.57239141  0.50728778  0.51896323
  0.69891196  0.66111323  0.64380596  0.19145411  0.49733934  0.32720705
  0.5488089   0.49649298  0.6157158   0.75780816  0.35492963  0.57446371
  0.32721105  0.63200183  0.53067634]
meanIOU: 0.550343
pixel acc: 0.896132
150.996609926s used to calculate IOU.

I did also only use standard pascal voc so perhaps the difference is as discussed in #4.

[aurora95]

Thanks @ahundt ! I've added the license. About the IOU, yes if you use the augmented pascal voc dataset you can get much higher IOU. Just note that you need to choose appropriate hyper-parameters.

[ahundt]

cool, thanks for the great tool. :-)

all items are addressed, closing