Normalize arXiv links

Currently, some of the links to arXiv go to the abstract page and some go
to the article PDF. Normalize it so that they all go to the abstract page.
You can get to the PDF from the abstract page, but you can't do the opposite,
so the abstract page is more useful.

CONTRIBUTING.md

@@ -150,7 +150,7 @@ description: >-
   This is a TensorFlow\* version of `densenet-121` model, one of the DenseNet
   group of models designed to perform image classification. The weights were converted
   from DenseNet-Keras Models. For details see repository <https://github.com/pudae/tensorflow-densenet/>,
-  paper <https://arxiv.org/pdf/1608.06993.pdf>
+  paper <https://arxiv.org/abs/1608.06993>
 task_type: classification
 files:
   - name: tf-densenet121.tar.gz

models/intel/human-pose-estimation-0001/description/human-pose-estimation-0001.md

@@ -21,7 +21,7 @@ The pose may contain up to 18 keypoints: ears, eyes, nose, neck, shoulders, elbo
 
 Average Precision metric described in [COCO Keypoint Evaluation site](http://cocodataset.org/#keypoints-eval).
 
-Tested on a COCO validation subset from the original paper [Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields](https://arxiv.org/pdf/1611.08050.pdf).
+Tested on a COCO validation subset from the original paper [Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields](https://arxiv.org/abs/1611.08050).
 
 ## Performance
 

models/intel/image-retrieval-0001/description/image-retrieval-0001.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Image retrieval model based on [MobileNetV2](https://arxiv.org/pdf/1801.04381.pdf) architecture as a backbone.
+Image retrieval model based on [MobileNetV2](https://arxiv.org/abs/1801.04381) architecture as a backbone.
 
 ## Example
 

models/intel/product-detection-0001/description/product-detection-0001.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-A product detector based on the SSD-lite architecture with [MobileNetV2](https://arxiv.org/pdf/1801.04381.pdf) as a backbone for self-checkout points of sale-related scenes.
+A product detector based on the SSD-lite architecture with [MobileNetV2](https://arxiv.org/abs/1801.04381) as a backbone for self-checkout points of sale-related scenes.
 The network can detect 12 classes of objects (`sprite`, `kool-aid`, `extra`, `ocelo`, `finish`, `mtn_dew`, `best_foods`, `gatorade`, `heinz`, `ruffles`, `pringles`, `del_monte`). Labels 0 and 1 are related to `background_label` and `undefined` correspondingly.
 
 ## Example

models/intel/single-image-super-resolution-1032/description/single-image-super-resolution-1032.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-[An Attention-Based Approach for Single Image Super Resolution](https://arxiv.org/pdf/1807.06779.pdf) but with reduced number of
+[An Attention-Based Approach for Single Image Super Resolution](https://arxiv.org/abs/1807.06779) but with reduced number of
 channels and changes in network achitecture. It enhances the resolution of the input image by a factor of 4.
 
 ## Example

models/intel/single-image-super-resolution-1033/description/single-image-super-resolution-1033.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-[An Attention-Based Approach for Single Image Super Resolution](https://arxiv.org/pdf/1807.06779.pdf) but with reduced number of
+[An Attention-Based Approach for Single Image Super Resolution](https://arxiv.org/abs/1807.06779) but with reduced number of
 channels and changes in network achitecture. It enhances the resolution of the input image by a factor of 3.
 
 ## Example

models/intel/text-detection-0003/description/text-detection-0003.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Text detector based on [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) architecture with [MobileNetV2-like](https://arxiv.org/pdf/1801.04381.pdf) as a backbone for indoor/outdoor scenes.
+Text detector based on [PixelLink](https://arxiv.org/abs/1801.01315) architecture with [MobileNetV2-like](https://arxiv.org/abs/1801.04381) as a backbone for indoor/outdoor scenes.
 
 ## Example
 
@@ -37,7 +37,7 @@ Text detector based on [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) archite
 
 2. name: "model/segm\_logits/add", shape: [1x2x192x320] - logits related to text/no-text classification for each pixel.
 
-Refer to [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) and demos for details.
+Refer to [PixelLink](https://arxiv.org/abs/1801.01315) and demos for details.
 
 ## Legal Information
 [*] Other names and brands may be claimed as the property of others.

models/intel/text-detection-0004/description/text-detection-0004.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Text detector based on [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) architecture with [MobileNetV2, depth_multiplier=1.4](https://arxiv.org/pdf/1801.04381.pdf) as a backbone for indoor/outdoor scenes.
+Text detector based on [PixelLink](https://arxiv.org/abs/1801.01315) architecture with [MobileNetV2, depth_multiplier=1.4](https://arxiv.org/abs/1801.04381) as a backbone for indoor/outdoor scenes.
 
 ## Example
 
@@ -37,7 +37,7 @@ Text detector based on [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) archite
 
 2. name: "model/segm\_logits/add", shape: [1x2x192x320] - logits related to text/no-text classification for each pixel.
 
-Refer to [PixelLink](https://arxiv.org/pdf/1801.01315.pdf) and demos for details.
+Refer to [PixelLink](https://arxiv.org/abs/1801.01315) and demos for details.
 
 ## Legal Information
 [*] Other names and brands may be claimed as the property of others.

models/intel/unet-camvid-onnx-0001/description/unet-camvid-onnx-0001.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-This is the UNet model that is designed to perform semantic segmentation. The model has been trained on the CamVid dataset from scratch using PyTorch framework. Training used median frequency balancing for class weighing. For details about the original floating point model, check out the [paper](https://arxiv.org/pdf/1505.04597.pdf).
+This is the UNet model that is designed to perform semantic segmentation. The model has been trained on the CamVid dataset from scratch using PyTorch framework. Training used median frequency balancing for class weighing. For details about the original floating point model, check out the [paper](https://arxiv.org/abs/1505.04597).
 
 The model input is a blob that consists of a single image of "1x3x368x480" in BGR order. The pixel values are integers in the [0, 255] range.
 

models/public/Sphereface/Sphereface.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-[Deep face recognition under open-set protocol](https://arxiv.org/pdf/1704.08063.pdf)
+[Deep face recognition under open-set protocol](https://arxiv.org/abs/1704.08063)
 
 ## Example
 

models/public/Sphereface/model.yml

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 description: >-
-  Deep face recognition under open-set protocol <https://arxiv.org/pdf/1704.08063.pdf>
+  Deep face recognition under open-set protocol <https://arxiv.org/abs/1704.08063>
 task_type: face_recognition
 files:
   - name: Sphereface.prototxt

models/public/colorization-v2-norebal/colorization-v2-norebal.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `colorization-v2-norebal` model is one of the [colorization](https://arxiv.org/pdf/1603.08511)
+The `colorization-v2-norebal` model is one of the [colorization](https://arxiv.org/abs/1603.08511)
 group of models designed to perform image colorization. For details
 about this family of models, check out the [repository](https://github.com/richzhang/colorization).
 
@@ -26,7 +26,7 @@ Model give as output predict A- and B-channels of LAB-image.
 ## Accuracy
 
 The accuracy metrics calculated on ImageNet
-validation dataset using [VGG16](https://arxiv.org/pdf/1409.1556.pdf) caffe
+validation dataset using [VGG16](https://arxiv.org/abs/1409.1556) caffe
 model and colorization as preprocessing.
 
 For preprocessing `rgb -> gray -> coloriaztion` recieved values:

models/public/colorization-v2/colorization-v2.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `colorization-v2` model is one of the [colorization](https://arxiv.org/pdf/1603.08511)
+The `colorization-v2` model is one of the [colorization](https://arxiv.org/abs/1603.08511)
 group of models designed to perform image colorization. For details
 about this family of models, check out the [repository](https://github.com/richzhang/colorization).
 
@@ -23,7 +23,7 @@ Model give as output predict A- and B-channels of LAB-image.
 ## Accuracy
 
 The accuracy metrics calculated on ImageNet
-validation dataset using [VGG16](https://arxiv.org/pdf/1409.1556.pdf) caffe
+validation dataset using [VGG16](https://arxiv.org/abs/1409.1556) caffe
 model and colorization as preprocessing.
 
 For preprocessing `rgb -> gray -> coloriaztion` recieved values:

models/public/ctpn/ctpn.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Detecting Text in Natural Image with Connectionist Text Proposal Network. For details see [paper](https://arxiv.org/pdf/1609.03605.pdf).
+Detecting Text in Natural Image with Connectionist Text Proposal Network. For details see [paper](https://arxiv.org/abs/1609.03605).
 
 ## Example
 

models/public/ctpn/model.yml

@@ -14,7 +14,7 @@
 
 description: >-
   Detecting Text in Natural Image with Connectionist Text Proposal Network. For
-  details see paper <https://arxiv.org/pdf/1609.03605.pdf>.
+  details see paper <https://arxiv.org/abs/1609.03605>.
 task_type: detection
 files:
   - name: ctpn.pb

models/public/deeplabv3/deeplabv3.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-DeepLab is a state-of-art deep learning model for semantic image segmentation. For details see [paper](https://arxiv.org/pdf/1706.05587.pdf).
+DeepLab is a state-of-art deep learning model for semantic image segmentation. For details see [paper](https://arxiv.org/abs/1706.05587).
 
 ## Example
 

models/public/deeplabv3/model.yml

@@ -14,7 +14,7 @@
 
 description: >-
   DeepLab is a state-of-art deep learning model for semantic image segmentation. For
-  details see paper <https://arxiv.org/pdf/1706.05587.pdf>.
+  details see paper <https://arxiv.org/abs/1706.05587>.
 task_type: semantic_segmentation
 files:
   - name: deeplabv3.tar.gz

models/public/densenet-121-caffe2/densenet-121-caffe2.md

@@ -6,7 +6,7 @@ This is a Caffe2\* version of `densenet-121` model, one of the DenseNet
 group of models designed to perform image classification. This model
 was converted from Caffe\* to Caffe2\* format. 
 For details see repository <https://github.com/caffe2/models/tree/master/densenet121>,
-paper <https://arxiv.org/pdf/1608.06993.pdf>.
+paper <https://arxiv.org/abs/1608.06993>.
 
 ## Example
 

models/public/densenet-121-caffe2/model.yml

@@ -16,7 +16,7 @@ description: >-
   This is a Caffe2* version of "densenet-121" model, one of the DenseNet group of
   models designed to perform image classification. This model was converted from Caffe*
   to Caffe2* format.  For details see repository <https://github.com/caffe2/models/tree/master/densenet121>,
-  paper <https://arxiv.org/pdf/1608.06993.pdf>.
+  paper <https://arxiv.org/abs/1608.06993>.
 task_type: classification
 files:
   - name: predict_net.pb

models/public/densenet-121-tf/densenet-121-tf.md

@@ -3,7 +3,7 @@
 ## Use Case and High-Level Description
 
 This is a TensorFlow\* version of `densenet-121` model, one of the DenseNet\*
-group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details, see [repository](https://github.com/pudae/tensorflow-densenet/) and [paper](https://arxiv.org/pdf/1608.06993.pdf).
+group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details, see [repository](https://github.com/pudae/tensorflow-densenet/) and [paper](https://arxiv.org/abs/1608.06993).
 
 ## Example
 

models/public/densenet-121-tf/model.yml

@@ -16,7 +16,7 @@ description: >-
   This is a TensorFlow* version of "densenet-121" model, one of the DenseNet* group
   of models designed to perform image classification. The weights were converted from
   DenseNet-Keras Models. For details, see repository <https://github.com/pudae/tensorflow-densenet/>
-  and paper <https://arxiv.org/pdf/1608.06993.pdf>.
+  and paper <https://arxiv.org/abs/1608.06993>.
 task_type: classification
 files:
   - name: tf-densenet121.tar.gz

models/public/densenet-121/densenet-121.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `densenet-121` model is one of the [DenseNet*](https://arxiv.org/pdf/1608.06993)
+The `densenet-121` model is one of the [DenseNet*](https://arxiv.org/abs/1608.06993)
 group of models designed to perform image classification. The authors originally trained the models 
 on Torch\*, but then converted them into Caffe\* format. All DenseNet models have
 been pretrained on the ImageNet image database. For details about this family of

models/public/densenet-121/model.yml

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 description: >-
-  The "densenet-121" model is one of the DenseNet* <https://arxiv.org/pdf/1608.06993>
+  The "densenet-121" model is one of the DenseNet* <https://arxiv.org/abs/1608.06993>
   group of models designed to perform image classification. The authors originally
   trained the models  on Torch*, but then converted them into Caffe* format. All DenseNet
   models have been pretrained on the ImageNet image database. For details about this

models/public/densenet-161-tf/densenet-161-tf.md

@@ -3,7 +3,7 @@
 ## Use Case and High-Level Description
 
 This is a TensorFlow\* version of `densenet-161` model, one of the DenseNet
-group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details see [repository](https://github.com/pudae/tensorflow-densenet/), [paper](https://arxiv.org/pdf/1608.06993.pdf).
+group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details see [repository](https://github.com/pudae/tensorflow-densenet/), [paper](https://arxiv.org/abs/1608.06993).
 
 ## Example
 

models/public/densenet-161-tf/model.yml

@@ -16,7 +16,7 @@ description: >-
   This is a TensorFlow* version of "densenet-161" model, one of the DenseNet group
   of models designed to perform image classification. The weights were converted from
   DenseNet-Keras Models. For details see repository <https://github.com/pudae/tensorflow-densenet/>,
-  paper <https://arxiv.org/pdf/1608.06993.pdf>.
+  paper <https://arxiv.org/abs/1608.06993>.
 task_type: classification
 files:
   - name: tf-densenet161.tar.gz

models/public/densenet-161/densenet-161.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `densenet-161` model is one of the [DenseNet](https://arxiv.org/pdf/1608.06993)
+The `densenet-161` model is one of the [DenseNet](https://arxiv.org/abs/1608.06993)
 group of models designed to perform image classification. The main difference with
 the `densenet-121` model is the size and accuracy of the model. The `densenet-161`
 is much larger at 100MB in size vs the `densenet-121` model's roughly 31MB size.

models/public/densenet-161/model.yml

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 description: >-
-  The "densenet-161" model is one of the DenseNet <https://arxiv.org/pdf/1608.06993>
+  The "densenet-161" model is one of the DenseNet <https://arxiv.org/abs/1608.06993>
   group of models designed to perform image classification. The main difference with
   the "densenet-121" model is the size and accuracy of the model. The "densenet-161"
   is much larger at 100MB in size vs the "densenet-121" model's roughly 31MB size.

models/public/densenet-169-tf/densenet-169-tf.md

@@ -3,7 +3,7 @@
 ## Use Case and High-Level Description
 
 This is a TensorFlow\* version of `densenet-169` model, one of the DenseNet
-group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details, see [repository](https://github.com/pudae/tensorflow-densenet/) and [paper](https://arxiv.org/pdf/1608.06993.pdf).
+group of models designed to perform image classification. The weights were converted from DenseNet-Keras Models. For details, see [repository](https://github.com/pudae/tensorflow-densenet/) and [paper](https://arxiv.org/abs/1608.06993).
 
 ## Example
 

models/public/densenet-169-tf/model.yml

@@ -16,7 +16,7 @@ description: >-
   This is a TensorFlow* version of "densenet-169" model, one of the DenseNet group
   of models designed to perform image classification. The weights were converted from
   DenseNet-Keras Models. For details, see repository <https://github.com/pudae/tensorflow-densenet/>
-  and paper <https://arxiv.org/pdf/1608.06993.pdf>.
+  and paper <https://arxiv.org/abs/1608.06993>.
 task_type: classification
 files:
   - name: tf-densenet169.tar.gz

models/public/densenet-169/densenet-169.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `densenet-169` model is one of the [DenseNet](https://arxiv.org/pdf/1608.06993)
+The `densenet-169` model is one of the [DenseNet](https://arxiv.org/abs/1608.06993)
 group of models designed to perform image classification. The main difference with
 the `densenet-121` model is the size and accuracy of the model. The `densenet-169`
 is larger at just about 55MB in size vs the `densenet-121` model's roughly 31MB size.

models/public/densenet-169/model.yml

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 description: >-
-  The "densenet-169" model is one of the DenseNet <https://arxiv.org/pdf/1608.06993>
+  The "densenet-169" model is one of the DenseNet <https://arxiv.org/abs/1608.06993>
   group of models designed to perform image classification. The main difference with
   the "densenet-121" model is the size and accuracy of the model. The "densenet-169"
   is larger at just about 55MB in size vs the "densenet-121" model's roughly 31MB

models/public/densenet-201/densenet-201.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-The `densenet-201` model is also one of the [DenseNet](https://arxiv.org/pdf/1608.06993)
+The `densenet-201` model is also one of the [DenseNet](https://arxiv.org/abs/1608.06993)
 group of models designed to perform image classification. The main difference with
 the `densenet-121` model is the size and accuracy of the model. The `densenet-201`
 is larger at over 77MB in size vs the `densenet-121` model's roughly 31MB size.

models/public/densenet-201/model.yml

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 description: >-
-  The "densenet-201" model is also one of the DenseNet <https://arxiv.org/pdf/1608.06993>
+  The "densenet-201" model is also one of the DenseNet <https://arxiv.org/abs/1608.06993>
   group of models designed to perform image classification. The main difference with
   the "densenet-121" model is the size and accuracy of the model. The "densenet-201"
   is larger at over 77MB in size vs the "densenet-121" model's roughly 31MB size.

models/public/facenet-20180408-102900/facenet-20180408-102900.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-FaceNet: A Unified Embedding for Face Recognition and Clustering. For details see the [repository](https://github.com/davidsandberg/facenet/), [paper](https://arxiv.org/pdf/1503.03832.pdf)
+FaceNet: A Unified Embedding for Face Recognition and Clustering. For details see the [repository](https://github.com/davidsandberg/facenet/), [paper](https://arxiv.org/abs/1503.03832)
 
 ## Example
 

models/public/facenet-20180408-102900/model.yml

@@ -14,7 +14,7 @@
 
 description: >-
   FaceNet: A Unified Embedding for Face Recognition and Clustering. For details see
-  the repository <https://github.com/davidsandberg/facenet/>, paper <https://arxiv.org/pdf/1503.03832.pdf>
+  the repository <https://github.com/davidsandberg/facenet/>, paper <https://arxiv.org/abs/1503.03832>
 task_type: face_recognition
 files:
   - name: 20180408-102900.zip

models/public/faster_rcnn_inception_resnet_v2_atrous_coco/faster_rcnn_inception_resnet_v2_atrous_coco.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Faster R-CNN with Inception Resnet v2 Atrous version. Used for object detection. For details see the [paper](https://arxiv.org/pdf/1506.01497v3.pdf).
+Faster R-CNN with Inception Resnet v2 Atrous version. Used for object detection. For details see the [paper](https://arxiv.org/abs/1506.01497v3).
 
 ## Example
 

models/public/faster_rcnn_inception_resnet_v2_atrous_coco/model.yml

@@ -14,7 +14,7 @@
 
 description: >-
   Faster R-CNN with Inception Resnet v2 Atrous version. Used for object detection.
-  For details see the paper <https://arxiv.org/pdf/1506.01497v3.pdf>.
+  For details see the paper <https://arxiv.org/abs/1506.01497v3>.
 task_type: detection
 files:
   - name: faster_rcnn_inception_resnet_v2_atrous_coco_2018_01_28.tar.gz

models/public/faster_rcnn_inception_v2_coco/faster_rcnn_inception_v2_coco.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Faster R-CNN with Inception v2. Used for object detection. For details, see the [paper](https://arxiv.org/pdf/1506.01497v3.pdf).
+Faster R-CNN with Inception v2. Used for object detection. For details, see the [paper](https://arxiv.org/abs/1506.01497v3).
 
 ## Example
 

models/public/faster_rcnn_inception_v2_coco/model.yml

@@ -14,7 +14,7 @@
 
 description: >-
   Faster R-CNN with Inception v2. Used for object detection. For details, see the
-  paper <https://arxiv.org/pdf/1506.01497v3.pdf>.
+  paper <https://arxiv.org/abs/1506.01497v3>.
 task_type: detection
 files:
   - name: faster_rcnn_inception_v2_coco.tar.gz

models/public/faster_rcnn_resnet101_coco/faster_rcnn_resnet101_coco.md

@@ -2,7 +2,7 @@
 
 ## Use Case and High-Level Description
 
-Faster R-CNN Resnet-101 model. Used for object detection. For details, see the [paper](https://arxiv.org/pdf/1506.01497v3.pdf).
+Faster R-CNN Resnet-101 model. Used for object detection. For details, see the [paper](https://arxiv.org/abs/1506.01497v3).
 
 ## Example