Pose Detection

Deep learning model for the classification of poses in images using Mask RCNN and VGG19. This model classifies 15 different poses. Model can be easily adapted to classify new poses (maximum of 2 persons) provided that the dataset is given. Outline of model architecture is here. Archive of submission for Today I Learned Ai Camp 2019, with a final validation score of 0.70989.

Model Architecture

Getting Started

Prerequisites

Clone this github respository

!git clone github.com/CannyAss/Pose-Detector

Installing

A step by step series of examples that tell you how to get a development env running

Say what the step will be

Give the example

And repeat

until finished

End with an example of getting some data out of the system or using it for a little demo

Running the tests

Explain how to run the automated tests for this system

Break down into end to end tests

Explain what these tests test and why

Give an example

And coding style tests

Explain what these tests test and why

Give an example

Built With

• Mask_RCNN - The library used for instance segmentation
• Keras - The deep learning module used for classification

Authors

• Chengyang - Generation of good masks from dataset
• Warren - Selection and curation of deep learning model
• Wei Kiat - A little bit of both
• Yuqing - Visualisation of pipeline and presentation

Acknowledgments

• Defence Science and Technology Agency for the dataset and competition experience