aneesahmedpro / rock-paper-scissor-vision

"Recognise Hand-Posture using Computer Vision." Make the machine look at the hand of a person using live camera feed and recognise the postures made by the hand (rock, paper or scissor) in real-time using a convolutional neural network.

Rock-Paper-Scissor Vision

Make the machine look at the hand of a person using live camera feed and recognise the postures made by the hand (rock, paper or scissor) in realtime using a convolutional neural network.

---

System Requirements

1. A camera device (e.g. the in-built webcam of laptop)
2. Python (3.6 or higher)
3. Numpy (1.13.3 or higher)
4. TensorFlow (1.8 or higher)
5. Either one of the following:
   1. PyGame (1.9.3 or higher)
   2. OpenCV (3.2.0 or higher)

PyGame vs OpenCV

To access the camera and to show the GUI windows on screen, a library is required which is capable of all that. This project currently supports 2 libraries: PyGame and OpenCV. Either one can be used.

To help you choose, here is a comparision:

OpenCV advantages:

• Works well on a variety of platforms, including Linux and Windows.

OpenCV disadvantages:

• Being a full-fledged computer vision library, it is too huge and heavy for a small task of accessing camera. The wheel is around 40 MB in size.
• Can be a huge hassle to install and get it working. But for many platforms it has been made a lot easier to setup, e.g. on Windows and on Ubuntu.

PyGame advantages:

• Being a light-weight game development library, the wheel file is small and only around 5 MB in size.

PyGame disadvantages:

• The camera accessing feature is (currently) experimental and maybe be removed in the future.
• The camera accessing feature currently only works on Linux platforms.

How to Install:

See the section "How to Install" in INSTRUCTIONS.md.