They did comparisons to other parkour games. The most exciting part must be that they force users to use their body languages to control direction or actions. It's way more healthier then those that only need to tap on the screen.

However, things exist for reasons. For me, I'm a lazy person, it's really hard to let me stand up and face the camera to actually jump. It's kind of not user-friendly in some way, especially for users like me. That is also why there are rare apps that are similar which is popular. The most similar one that I thought of is PokemonGo, which need you to go around. Sorry to point that out, which is only my personal thought.

By experience the full product, they showed a extraordinary great ability of research, investigation, analysis and design skills.
The game has a high degree of completion. They even combine all things together as a .exe file.
The control delay is not even under awareness.
The game is overall very interesting. It has coins to collect, monsters to avoid and even has a huge gift waiting at the end. I fell great achivement when reaching it.
3D model looks great and natual, although they only used free resources.
Their face detection module is written in python. The game is in C#. They connect works great with Socket protocol

They tested the game with 62 test cases, which works great.
As for their github, I didn't find automatic testes documents.
I tried their game together with their face detection.

Everything looks great as a console game.
But it would be better to have online functions to show user's performance statistic, best record or provide ranking list, etc. For that, they may need database to store all these date. Also they need Authtication to login.

One great idea is to allow users to unlock unique abilities with coins they get. It may includes higher jump, double the coins they get, etc. Just giving a idea.

1. great code management.
2. Detection part is accurate.
3. The demo video shows fluent flow of your game.
4. I think the process seems like more fluent than last time.
   As a player can I change the speed of running while running? And I can add some gears to the character? To make it more options.
   Also, if you can improve the settings of the environment, it will be more fun.

Face Detection. Met this problem when trying to run.

OpenCV Error: Assertion failed (scn == 3 || scn == 4) in cvtColor, file /build/opencv-RI6cfE/opencv-2.4.9.1+dfsg1/modules/imgproc/src/color.cpp, line 3737
Traceback (most recent call last):
  File "detectFace.py", line 31, in <module>
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.error: /build/opencv-RI6cfE/opencv-2.4.9.1+dfsg1/modules/imgproc/src/color.cpp:3737: error: (-215) scn == 3 || scn == 4 in function cvtColor

After looking through your code, you actually have done a very good work at using opencv to detect face. The overall structure is good. Unfortunately, I can't experience the actual performance of your detection module.

The game part. Never used Unity3d before, so it's a bit difficult to ascertain where what goes. Sorry that I can't give any comments. Please update README file to show how to run your code with the exact running environment and instrutions.

But though the video, you guys seems like achieved a very amazing progress of your project. It looks so interesting that I really want to give it a try.

One question is I don't understand how the face detection module interact with your game. Does it run inside your game? Please add some information about your work if you can.
Looking forward for your final project!

Besides the project structure question, I think I will focus more on the quality of the codes.
First, I have run the Face detection code of the program.
It seems working well in the window.But when the face is around the corner of the window, the time delay seems a little long. And the name of the output value is kind of confusing, I mean, I do not know the size of the window and where our face is in the windows, maybe you could output the shape of the window, too.
Second, I planed to run the Unity-3D part, however, there is no Unity-3D on my computer. So, I can only check the sample video, it works pretty well in the Video. The moving of the objects in the video is simultaneous with the moving of the hand. And the simulation of the object collision is also great.
All in all, I think the function of the program is great, the simulation part is still black box to me.
Guys, you've done great job.

Readme file is not clear. So i am not sure about your project structure.
Code management seems like good.
Facedection part seems like good.
Demo video looks good.
The Interface can be more colorful and more playful.
Maybe user can switch to different mode ?