aaronmarkham commented on a change in pull request #13647: Update lip reading 
example
URL: https://github.com/apache/incubator-mxnet/pull/13647#discussion_r246879617
 
 

 ##########
 File path: example/gluon/lipnet/README.md
 ##########
 @@ -0,0 +1,115 @@
+# LipNet: End-to-End Sentence-level Lipreading
+
+---
+
+Gluon implementation of [LipNet: End-to-End Sentence-level 
Lipreading](https://arxiv.org/abs/1611.01599)
+
+![net_structure](asset/network_structure.png)
+
+## Requirements
+- Python 3.6.4
+- MXnet 1.3.0
+- The Required Disk Space: 35Gb
+```
+pip install -r requirements.txt
+```
+
+## Test Environment
+- 4 CPU cores
+- 1 GPU (Tesla K80 12GB)
+
+
+## The Data
+- The GRID audiovisual sentence corpus 
(http://spandh.dcs.shef.ac.uk/gridcorpus/)
+- Video: (normal)(480 M each)
+- Align: word alignments(190 K each) 
+
+## Prepare the Data
+### Download the data
+- arguments
+  - src_path : Path for videos (default='./data/mp4s/')
+  - align_path : Path for aligns (default='./data/')
+  - n_process : num of process (default=1)
+
+```
+cd ./utils && python download_data.py --n_process $(nproc)
+```
+
+### Preprocess the Data: Extracting the mouth images from a video and save it.
+- arguments
 
 Review comment:
   @thomelane I had some issues earlier, but they seem to be resolved now. The 
files are being sync'd to s3://mxnet-public/lipnet/ - they're not done 
uploading yet.
   
   It has:
   lipnet/data/
   lipnet/data/align/  #with a bunch of .align files
   lipnet/data/datasets/ #with a bunch of subfolders containing .png files (of 
mouths)
   
   Like this:
   
![mouth-preprocessed-dataset-example](https://s3.us-east-2.amazonaws.com/mxnet-public/lipnet/data/datasets/s29/brijzn/mouth_069.png)
   
   So... the readme could have the following helper for getting the 
preprocessed data.
   From the lipnet project root:
   ```
   aws s3 sync s3://mxnet-public/lipnet/ .
   ```
   I've tested the above command and it is busy syncing right now on another 
instance!

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to