This project is in reference to the Hacker Earth Deep Learning Challenge Identify the dance form
The classifier uses cnn to identify the image of a dance form.
Following are the class categories for identification:
- Manipuri
- Bharatanatyam
- Odissi
- Kathakali
- Kathak
- Sattriya
- Kuchipudi
- Mohiniyattam
The dataset for the same is available at: Data
Make sure your python enviroment fullfils the following deps. If not, install the following using 'pip install ___dep___'
Python Dependencies:
- Open Cv
- Pandas
- Tensorflow
- Numpy
After downloading the Dataset, place it in a folder named: data and run preprocess.py. When the preprocessing is finnished, you should have two new files with the name: features.py and labels.npy. Now run the model.py which will save the model to cnn.model dir. You are now ready to make predictions. To test an image, place it in the data/test/ folder and run predict.py.. Prediction results are saved in predictions.csv file