Molecular Playground

Local Python Server

• Adapted from Victor Dibia, Real-time Hand-Detection using Neural Networks (SSD) on Tensorflow, (2017), GitHub repository, https://github.com/victordibia/handtracking

To Run

Make sure that you have python >3.6 installed ([install here], as well as Java.(https://www.python.org/downloads/))

1. Clone the repository

git clone <url>

2. Run the detection script (Note: this should take several seconds, as the NN graph is loaded and the workers begin)

python detect_mult_threaded.py

1. Run the local JMOL molecule.

java -jar MPJmolApp.jar

---

Script Options

• -src : video source, default = 0
• -nhands : number of hands, default = 2
• -fps : show frames per second: default = 1 [yes]
• -display : show output for display: default = 1 [yes]
• -num-w : number of workers for processing: default = 4
• -flip-x-axis: flip x axis if camera facing installation: default = 0 [no]
• -flip-y-axis: flip y axis if camera facing installation: default = 0 [no]

example: python detect_multi_threaded.py -nhands 1 -fps 1 -ds 1 -num-w 5