Add readmes to src and aws_scripts

Signed-off-by: Cody W. Eilar <[email protected]>

src/README.md

@@ -0,0 +1,35 @@
+# SRC directory
+
+This directory contains many different source codes for achieving the `extract_features`
+pipeline. The C++ code is the workhorse of the entire project,
+the MATLAB code is the inspiration for the C++ code, most of
+the python code is for running AWS services, and finally the R
+code is used for classifying the output of the C++ code.
+Below is a breif description of what is contained in
+each of the directories:
+
+- *aws_scripts* - This is all python code for running
+`extract_features` on the cloud.
+- *datatypes* - A small collection of C++ objects used for
+general purpose data strucutures
+- *experiments* - Most of the code in here is for bench
+marking various aspects of the extract features program.
+There are speed comparisons between using `Mat` and `UMat`
+datatypes, as well as comparisons for different implementations
+of overlap and add.
+- *ipython_notebooks* - Any python notebooks used for
+generating plots related to `extract_features`.
+- *matlab* - The original code used validate that
+classification of activities is possible using optical flow
+- *optical_flow* - This is where the majority of the C++ library code lives
+to extract features using optical flow. There are several
+implementations inside the directory. All of the classes are
+templatized.
+- *programs* - There are a few misellaneous programs inside
+this directory, but it also contains the main program of `extract_features`.
+- *r_code* - This is where the classifier R code resides.
+- *readers* - Originally, the intent of this directory
+was to contain several types of video readers. I.e. I wanted
+a way to read video formats but I also wanted to be able to read data stored in an hdf5 file. The HDF5 code was never implemented.
+- *utilities* - A library of useful utilities for working with videos for this project. Mostly they are related
+to *OpenCV* functions. 

src/aws_scripts/README.md

@@ -0,0 +1,35 @@
+# AWS Scripts
+
+In order to distribute the `extract_features` program on the cloud,
+code in this directory must be leveraged. In order to
+successfully run this code, there are a few prerequisites
+that must be first met:
+
+## Prerequisites
+- Must have an AWS account
+- Must have access to the S3 bucket hosting the AOLME videos
+(Please email [Cody Eilar](mailto:[email protected]) if you wish to have access)
+- Must have an ECS cluster setup that pulls from the latest
+optical flow docker build located here: [AWS Container Registry](883856218903.dkr.ecr.us-west-2.amazonaws.com/optical-flow:latest).
+ - `docker pull 883856218903.dkr.ecr.us-west-2.amazonaws.com/optical-flow:latest`
+ - If you have permission denied, be sure you follow
+ the steps on the AWS site for performing a docker pull,
+ and then you made need to request access from [Cody Eilar](mailto:[email protected]).
+- Once the cluster is properly configured, you will then
+be able to run the scripts in this directory.
+
+## Description of Scripts
+- `add_videos_to_process.py` This is the script that is used
+to add video uris to an SQS queue. The videos are then popped
+off the queue by the running docker images and the output
+of extract_features is put into another queue, which is then read by this program and stored into a variable.
+- `message_type.py` - This contains the layout of the messages
+sent to the SQS queue for processing.
+- `process_video_list.py` - This is ultimately the script
+that does all the processing in the Docker image. It essentially
+pulls video URI's off the SQS queue and processes them. When
+processing is done, it then puts the answer onto another SQS
+queue to be processed by the master node.
+- `example_videos_to_process.txt` - This is an example input
+into the `add_videos_to_process.py` script. This
+illustrates how to send videos to the processing nodes.