2014 RIT Department of Electrical and Microelectronic Engineering
Artificial Intelligence project - Conducted by Jordan O'Connor (jmo3837@rit.edu) and Bryan Beatrez (bpb4868@rit.edu) for AI Explorations (EEEE54701.2141) at the Rochester Institute of Technology.
Much of this code is adopted from the Cornell Robot Learning Lab. Computer Science Department, Cornell University. http://pr.cs.cornell.edu/humanactivities/index.php. The specific data used is Person 1 - 4 from the CAD-60 dataset.
##The Big Picture This project is an idea to combine two very well established AI topics, motion classification and music classification, and combine them to create an system that selects a genre/playlist based on the mood classification of the environment surrounding it.
The big picture would be this: Design an app for the XBox One, which is useful because it already has the imaging tools and music database, that would analyze a room for the user and play music based on what it sees and other parameters (like time of day, day of the week, weather outside). As you would think, a different playlist would be selected at a Friday night college party versus a quiet and rainy tuesday morning while you are reading a book (or surfing the web).
The work completed for this project focused more on the classifaction of movement in a environment.
Included in this folder is code to generate our classified training data, apply our algorithms and calculate our results. https://github.com/Ohhhhhconnor/AI_Project/tree/master/project_files
Included in this folder is the CSV files needed to generate the joint skeleton, which is used for the training data in the Project Files. https://github.com/Ohhhhhconnor/AI_Project/tree/master/video_data
- Package Needed: Statistics Toolbox
- Only tested on MATLAB(R) 2014b
- Download the project zip files from the github page: https://github.com/Ohhhhhconnor/AI_Project
- In MATLAB, add the project files to the path.
- Update the file path in project_files >> ApplyTraining >> apply_training_data.m to match the absolute path of the CSV file called training_data.txt.
- To generate our current results, run the apply_training_data.m script in MATLAB.
- If you would like to edit the number of frames in the training data, the variable frames_per_video should be edited in the files apply_training_data.m, generate_training_data.m and generate_classification_array.m.
- To use more frames, after updating the frames_per_video variable, run the generate_training_data.m script and allow it to finish. The generated file will be indicated at the end of the script (the path can be edited at the end of generate_training_data.m).
- Run apply_training_data.m to generate the new results.
2014 RIT Department of Electrical and Microelectronic Engineering