add i3d_slow to mxnet (#1568)

* add i3d_slow to mxnet

* lint and add k700 classses

* add readme

* readme

Author: bryanyzhu

README.md

@@ -46,7 +46,7 @@ Check the HD video at [Youtube](https://www.youtube.com/watch?v=nfpouVAzXt0) or
 | [Semantic Segmentation:](https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation) <br/>associate each pixel of an image <br/> with a categorical label. |  <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation"><img src="docs/_static/semantic-segmentation.png" alt="semantic" height="200"/></a> | <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">FCN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">PSP</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">ICNet</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">DeepLab-v3</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">DeepLab-v3+</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">DANet</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#semantic-segmentation">FastSCNN</a> |
 | [Instance Segmentation:](https://gluon-cv.mxnet.io/model_zoo/segmentation.html#instance-segmentation) <br/>detect objects and associate <br/> each pixel inside object area with an <br/> instance label. | <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#instance-segmentation"><img src="docs/_static/instance-segmentation.png" alt="instance" height="200"/></a> | <a href="https://gluon-cv.mxnet.io/model_zoo/segmentation.html#instance-segmentation">Mask RCNN</a>|
 | [Pose Estimation:](https://gluon-cv.mxnet.io/model_zoo/pose.html) <br/>detect human pose <br/> from images. | <a href="https://gluon-cv.mxnet.io/model_zoo/pose.html"><img src="docs/_static/pose-estimation.svg" alt="pose" height="200"/></a> | <a href="https://gluon-cv.mxnet.io/model_zoo/pose.html#simple-pose-with-resnet">Simple Pose</a>|
-| [Video Action Recognition:](https://gluon-cv.mxnet.io/model_zoo/action_recognition.html) <br/>recognize human actions <br/> in a video. | <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html"><img src="docs/_static/action-recognition.png" alt="action_recognition" height="200"/></a> | MXNet: <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">C3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">P3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R2+1D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">Non-local</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">SlowFast</a> <br/> PyTorch: <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D_slow</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R2+1D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">Non-local</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">CSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/SlowFast.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TPN</a> |
+| [Video Action Recognition:](https://gluon-cv.mxnet.io/model_zoo/action_recognition.html) <br/>recognize human actions <br/> in a video. | <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html"><img src="docs/_static/action-recognition.png" alt="action_recognition" height="200"/></a> | MXNet: <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">C3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D_slow</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">P3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R2+1D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">Non-local</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">SlowFast</a> <br/> PyTorch: <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">I3D_slow</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">R2+1D</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">Non-local</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">CSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/SlowFast.html">TSN</a>, <a href="https://gluon-cv.mxnet.io/model_zoo/action_recognition.html">TPN</a> |
 | [Depth Prediction:](https://gluon-cv.mxnet.io/model_zoo/depth.html) <br/>predict depth map <br/> from images. | <a href="https://gluon-cv.mxnet.io/model_zoo/depth.html"><img src="docs/_static/depth.png" alt="depth" height="200"/></a> | <a href="https://gluon-cv.mxnet.io/model_zoo/depth.html#kitti-dataset">Monodepth2</a>|
 | [GAN:](https://github.com/dmlc/gluon-cv/tree/master/scripts/gan) <br/>generate visually deceptive images | <a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/gan"><img src="https://github.com/dmlc/gluon-cv/raw/master/scripts/gan/wgan/fake_samples_400000.png" alt="lsun" height="200"/></a> | <a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/gan/wgan">WGAN</a>, <a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/gan/cycle_gan">CycleGAN</a>, <a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/gan/stylegan">StyleGAN</a>|
 | [Person Re-ID:](https://github.com/dmlc/gluon-cv/tree/master/scripts/re-id/baseline) <br/>re-identify pedestrians across scenes | <a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/re-id/baseline"><img src="https://user-images.githubusercontent.com/3307514/46702937-f4311800-cbd9-11e8-8eeb-c945ec5643fb.png" alt="re-id" height="160"/></a> |<a href="https://github.com/dmlc/gluon-cv/tree/master/scripts/re-id/baseline">Market1501 baseline</a> |

docs/model_zoo/action_recognition_mxnet.rst

@@ -68,6 +68,19 @@ The following table lists pre-trained models trained on Kinetics400.
    :class: tight-table
    :widths: 30 12 10 10 8 10 12 8
 
+
+Kinetics700 Dataset
+-------------------
+
+The following table lists our trained models on Kinetics700.
+
+.. csv-table::
+   :file: ./csv_tables/Action_Recognitions/Kinetics700.csv
+   :header-rows: 1
+   :class: tight-table
+   :widths: 30 12 10 10 8 10 12 8
+
+
 UCF101 Dataset
 --------------
 

docs/model_zoo/action_recognition_torch.rst

@@ -29,6 +29,8 @@ The following table lists our trained models on Kinetics400.
 
   ``Segment`` is the number of segments used during training. For testing (reporting these numbers), we use 250 views for 2D networks (25 frames and 10-crop) and 30 views for 3D networks (10 clips and 3-crop) following the convention.
 
+  The model weights of ``r2plus1d_v2_resnet152_kinetics400``, ``ircsn_v2_resnet152_f32s2_kinetics400`` and ``TPN family`` are ported from VMZ and TPN repository. You may ignore the training config of these models for now.
+
 
 .. csv-table::
    :file: ./csv_tables/Action_Recognitions/Kinetics400_torch.csv

docs/model_zoo/csv_tables/Action_Recognitions/Kinetics700.csv

@@ -0,0 +1,2 @@
+Name,Pretrained,Segments,Clip Length,Top-1,Hashtag,Train Command,Train Log
+i3d_slow_resnet101_f16s4_kinetics700 [8]_,Scratch,1,16 (64/4),67.65,299b1d9d,NA,NA

gluoncv/data/__init__.py

@@ -23,6 +23,7 @@
 from .mixup.detection import MixupDetection
 from .ucf101.classification import UCF101, UCF101Attr
 from .kinetics400.classification import Kinetics400, Kinetics400Attr
+from .kinetics700.classification import Kinetics700, Kinetics700Attr
 from .somethingsomethingv2.classification import SomethingSomethingV2, SomethingSomethingV2Attr
 from .hmdb51.classification import HMDB51, HMDB51Attr
 from .video_custom.classification import VideoClsCustom

gluoncv/data/kinetics700/__init__.py

@@ -0,0 +1,6 @@
+# pylint: disable=wildcard-import
+"""Video action recognition, Kinetics700 dataset.
+https://deepmind.com/research/open-source/open-source-datasets/kinetics/
+"""
+from __future__ import absolute_import
+from .classification import *