Skip to content

Commit

Permalink
Bug fixes for style transfer model (onnx#193)
Browse files Browse the repository at this point in the history
* Create image

* Delete image

* Creating Image folder

* created image_classificatioin folder

Also added MobileNet to the folder

* delete mobilenet file

* Moving MobileNet README.md

* moving moblenet readme to the right folder

* creating image_classification folder

* Moving train_mobilenet.ipynb to MobileNet

* delete test.md

* Delete Test.md

* Create test.md

* Create test.md

* Update readme.md

changed file path for MobileNet

* Moving ResNet README.md to image_classification

* Update README

changed the file path for mobilenet

* Moving SqueezeNet ReadMe into image classification

* Move README.md

* Moving train_resnet.ipynb to ResNet

* Moved SqueezeNet train into image classification

* Move tiny_yolov2

* Updated ResNet destination

* Moved SSD

* Added test.md

* Moving LICENSE to AlexNet

* Moved SSD to the correct folder

* Moved VGG ReadMe to image classification

* Moving README.md to AlexNet

* Updated AlexNet destination

* Moved Faster-RCNN README.md

* Moved VGG train into image classification

* Updated models destination

* moved coco_classes.txt

* Moved coco_classes.txt

* Moved LICENSE to CaffeNet

* Put GoogleNet ReadMe into Inception/GoogleNet folder

* Moved README.md to CaffeNet

* Updated CaffeNet destination

* Moved License into Inception/googlenet folder

* Moved LICENSE to RCNN_ILSVRC13

* Moved Inception ReadMe to inception/googlenet folder

* Moved README.md to RCNN_ILSVRC13

* Moved Inception_V2 README to inception/googlenet folder

* Moved README.md to DenseNet-121

* Moved ShuffleNet README to image classification folder

* Moved README.md to ZFNET-512

* moved demo.jpg

* Moved README.md to MNIST

* Delete original demo.jpg

* moved mask_rcnn readme.md

* moved mask_rcnn coco_classes.txt

* Updated model destination

* Uploaded demo.jpg to mask-rcnn

* Delete demo.jpg

* Moved README.md for YOLOv3

* Updated models destination

* Updated model name

* Moved DUC README.md

* Updated Model name

* moved cityscapes_labels.py

* Updated models destination

* Moved cityscapes_loader.py

* Updated MNIST model name

* Update MNIST model name

* moved duc-postprocess.py

* moved duc-preprocess.py

* moved utils.py

* Moved README.md to ArcFace

* Moved arcface_inference.ipynb to ArcFace

* Moved arcface_validation.ipynb to ArcFace

* Moved face_image.py to ArcFace

* Update Hand file path

* Moved face_postprocess.py to ArcFace

* Moved README.md to Body folder

* Moved face_postprocess.py to body folder

* Moved EmotionFerplus to body/face/gesture folder

* Moved arcface_inference.ipynb to body folder

* Moved arcface_validation.ipynb to body folder

* Moved face_image.py to body folder

* Moved face_preprocess.py to ArcFace

* moved duc inference file

* Moved model destination

* Moved fresnet.py to ArcFace

* moved duc-validation file

* Moved ArcFace into Body folder

* moved duc inference and validation

* Moved destination of verification.py of ArcFace

Verfication.py of ArcFace put in body/face/gesture anaylsis folder

* finished changing folder structure

* changed folder links in README.md

* added link to bidir in README.md

* deleted text.md files

* decapitalized file names and undid line ending changes

* Testing out decapitalizing folder names

* decapitalized all folder names

* Reverted README.md to the original

* updated file folder links in README.md and shortened folder names

* Updated ArcFace file path in README.md

* Fixed broken links in the main README.md and changed folder names (vision & mnist)

* Fixed ImageNet links

* moved imagenet contents up a level to fix broken llinks

* moved imagenet contents up a level to fix broken links

* reverted line ending change

* reverted line ending changes in extact_imagenet.py

* testing line ending change in extact_imagenet.py

* testing line ending change in extact_imagenet.py

* Update extact_imagenet.py

* fixed line ending changes in extact_imagenet.py

* testing merge conflict

* deleted README.md_HEAD

* added empty README.md and added all images

* added onnx models in a folder

* Added lfs to onnx files

* added models with sample test data in .tar.gz format

* added fns-pytorch-to-onnx.ipynb

* finished modifying fns-pytorch-to-onnx.ipynb

* finished README.md and added onnx_helper.py

* Fixed table format and added amber.jpg to README

* Centered input image and scaled it down

* deleted .gitattributes

* deleted onnx_helper.py and removed 1st cell in style-transfer-ort.ipynb

* deleted onnx_helper.py and deleted 1st cell in style-transfer-ort.ipynb

* deleted .gitattributes

* reconverted the models to change input shape to (1, 3, 224, 224); updated all .gz files

* fixed broken links in README and updated size of .gz files

* changed output shape from [10, 3, H, W] to [1, 3, H, W] in README

* added .gitattributes

* deleted line space changes in .gitattributes

* added new line at end of .gitattributes file

* changed ) to ] in README

* added license information

* changed model sizes from kb to mb and onnx version to 1.4 in readme

* clarified what scale means in readme and jupyter notebook

* updated README title

* got rid of duplicated images in README

* deleted images folder

* fixed broken links in README.md

* changed model files from regular git to git LFS

* added style transfer link to main README

* moved models to style_transfer --> fast_neural_style subfolder

* deleted last sentence of fns description in README
jennifererwangg authored and prasanthpul committed Aug 16, 2019
1 parent e643d26 commit 8450a58
Showing 23 changed files with 33 additions and 2 deletions.
1 change: 1 addition & 0 deletions README.md
Original file line number Diff line number Diff line change
@@ -85,6 +85,7 @@ Image manipulation models use neural networks to transform input images to modif
|-|-|-|
|Unpaired Image to Image Translation using Cycle consistent Adversarial Network|[Zhu et al.](https://arxiv.org/abs/1703.10593)|The model uses learning to translate an image from a source domain X to a target domain Y in the absence of paired examples. <br>[contribute](contribute.md)|
|Image Super resolution using deep convolutional networks | [Dong et al.](http://ieeexplore.ieee.org/document/7115171/?reload=true) |A deep CNN that takes low-resolution image as the input and outputs the high-resolution one. Fast speed for restoration quality. <br>[contribute](contribute.md)|
|[Fast Neural Style Transfer](vision/style_transfer) | [Johnson et al.](https://arxiv.org/abs/1603.08155) |This method uses a loss network pretrained for image classification to define perceptual loss functions that measure perceptual differences in content and style between images. The loss network remains fixed during the training process.|
<hr>

### Speech & Audio Processing <a name="speech"/>
Original file line number Diff line number Diff line change
@@ -19,7 +19,7 @@ The model uses the method described in [Perceptual Losses for Real-Time Style Tr
<hr>

## Inference
Refer to [fns-pytorch-to-onnx.ipynb](fns-pytorch-to-onnx.ipynb) for detailed preprocessing and postprocessing.
Refer to [style-transfer-ort.ipynb](style-transfer-ort.ipynb) for detailed preprocessing and postprocessing.

### Input to model
The input to the model are 3-channel RGB images. The images have to be loaded in a range of [0, 255]. If running into memory issues, try resizing the image by increasing the scale number.
@@ -56,7 +56,7 @@ The original fast neural style model is from [pytorch/examples/fast_neural_style
<hr>

## Training
Refer to [pytorch/examples/fast_neural_style](https://github.com/pytorch/examples/tree/master/fast_neural_style) for training details in PyTorch. Refer to [fns-pytorch-to-onnx.ipynb](fns-pytorch-to-onnx.ipynb) for the conversion process.
Refer to [pytorch/examples/fast_neural_style](https://github.com/pytorch/examples/tree/master/fast_neural_style) for training details in PyTorch.
<hr>


3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/candy.onnx
Git LFS file not shown
3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/candy.tar.gz
Git LFS file not shown
3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/mosaic.onnx
Git LFS file not shown
3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/mosaic.tar.gz
Git LFS file not shown
Git LFS file not shown
Git LFS file not shown
Git LFS file not shown
Git LFS file not shown
3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/udnie.onnx
Git LFS file not shown
3 changes: 3 additions & 0 deletions vision/style_transfer/fast_neural_style/models/udnie.tar.gz
Git LFS file not shown
Binary file removed vision/style_transfer/models/candy.onnx
Binary file not shown.
Binary file removed vision/style_transfer/models/candy.tar.gz
Binary file not shown.
Binary file removed vision/style_transfer/models/mosaic.onnx
Binary file not shown.
Binary file removed vision/style_transfer/models/mosaic.tar.gz
Binary file not shown.
Binary file removed vision/style_transfer/models/pointilism.onnx
Binary file not shown.
Binary file removed vision/style_transfer/models/pointilism.tar.gz
Binary file not shown.
Binary file removed vision/style_transfer/models/rain_princess.onnx
Binary file not shown.
Binary file not shown.
Binary file removed vision/style_transfer/models/udnie.onnx
Binary file not shown.
Binary file removed vision/style_transfer/models/udnie.tar.gz
Binary file not shown.

0 comments on commit 8450a58

Please sign in to comment.