Software and Machine Learning Project: Dectect "Hailing a cab" motion.

Task

You and your partner must implement a system which, given a video stream from your webcam, detects a person who is hailing a cab. I.E if either hand is above the head, we should print "Someone is hailing a taxi!".

All code edits should occur in src/assignment.py.

This assignment requires you to install external libraries in order to run someone else's code (which estimates 'pose'). To do that you must follow the instructions below.

Once you have assignment.py running you must write your own python to detect when a taxi is hailed.

**hint:** when you see a code snippet you should run this command in your command prompt

1. Install and Run the Software

Get the Repo

You did this when you created your me repo, see the Lab 1 slides.

1. Fork this github repo
2. Clone your forked repo
3. Change directory into your newly created repo cd tf-pose-estimation

Create an environment

In order to run the software you must install the dependencies in an Anaconda Environment:
conda create -n <NAME-YOUR-ENV> python=3.7
once you've created your environment you need to activate it:
conda activate <NAME-YOUR-ENV>
If your environment is active you should see the name of your env in the command prompt
i.e.

(code-assgn1) C:/<YOUR_PATH>/tf-pose-estimation $

Install Dependencies

Now you've created and activated your environment you need to install some import dependencies. Type the following in your command prompt:

pip install -r requirements.txt
pip install tensorflow
conda install -c menpo opencv

You should now have all the dependencies to start the assignment!

Test Inference (one the dependencies are installed)

You can test the inference feature by capturing video from your webcam.

python src/assignment.py

Then you will see something like below, but with your own camera.

Hail a Taxi

Edit assignment.py in the src folder so that you detect a taxi being hailed (any of your arms are raised)
Read the code and TODOs in assignment.py. Uncomment the line:

print([(POSE_COCO_BODY_PARTS[k], v.x, v.y) for k,v in human.body_parts.items()])

and run it to see what happens.

Marking

In the following week you and your partner will present your cab hailing for marking by the end of the lesson. You must also explain the logic of your code to your marker.

As this is a paired assignment, you must also learn how to code in a team using git with your partner.
You and your partner will need to commit to your tf-pose-estimation repo at least once.
You must also co-author a commit with your partner.

Submission

Create, commit, and push a file named results.json. To your me\week 6 directory. This is a dictionary like file which contains:

1. your github username
2. your partners github username
3. a link to one of your commits
4. a link to one of your partners commits
5. a link to your co-authored commits
6. a screen shot of you cab hailing
7. a screen shot of your partner cab hailing

The results.json file should be commited and pushed as follows:

{
  "username_me": "YOURNAME",
  "username_partner": "THEIRNAME",
  "commit_me": "https://github.com/<YOUR-GITHUB>/tf-pose-estimation/commit/<thecommitSHA>",
  "commit_partner": "https://github.com/<YOUR-GITHUB>/tf-pose-estimation/commit/<thecommitSHA>",
  "commit_coauthor": "https://github.com/<YOUR-GITHUB>/tf-pose-estimation/commit/<thecommitSHA>",
  "screenshot_me": "/path/to/your/screenshot.jpg",
  "screenshot_partner": "/path/to/partner/screenshot.jpg"
}

Remember this goes in your me\week 6 directory.

Deadline

19th of July 2019. Marking will be done in the labs on the day. You must commit and push your results BEFORE the end of the lesson (2pm).