Project on Pattern Detection and Recognition using Deep Learning

By Manasi Khapke, Shraddha Gunjal, Faizan Shaikh

Introduction

Our main objective of doing this project is to understand the Deep Learning, more specifically Convolutional Neural Networks (CNN) for computer vision and use them to solve real world problems.

We demonstrate the use of CNN for two tasks, Detection and Recognition.

Code files description

• toy_classifier.ipynb : Digit Recognition for MNIST

• toy_localizer.ipynb : Facial Keypoints Detection

• whale_localizer.ipynb : Right Whale Localizer

• whale_classifier.ipynb : Right Whale Classifier

• INC_POSTER.pdf : Our poster for the project

Acknowledgement

Thanks to

• NOAA Fisheries for the excellent competition

• Kaggle for giving us the chance to test our algorithms

• PICT college for providing guidance and hardware support.

Special thanks to

• The staff of CS231n for giving insights to explore the fascinating field of Deep Learning.

• our family and friends, without their help and support this would never have been achieved.

References

• Honari, Sina, et al. "Recombinator Networks: Learning Coarse-to-Fine Feature Aggregation." arXiv preprint arXiv:1511.07356 (2015).

• Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural networks." Advances in neural information processing systems. 2012.

• https://github.com/ebenolson/kaggle-whale

• https://github.com/anlthms/whale-2015

• http://danielnouri.org/notes/2014/12/17/using-convolutional-neural-nets-to-detect-facial-keypoints-tutorial/

• http://cs231n.github.io/