/ITS-Project-9786-PG

Turn video files into images, using Haar cascades face detection crop faces from images. Run the images through a pre-trained GoogLeNet.

Primary LanguagePython

20S2-29: Depression Analysis of Facial Video Data using Deep Learning

Unit 9786 PG: ITS Capstone Project - University of Canberra

ITS project 29-S2: Depression Analysis from Facial Video Data via Deep Learning

Date submitted: 6 NOV 2020

Date last updated: 6 NOV 2020

Project team members: Hang Hoang - u3197442 Charmane Foo - u3201698 Matt Lally - u3167761 Lakmal Attanayake - u3177896

This repository contains code for depression analysis of facial video data using pre-trained deep learning networks.

Turn video files into images, using Haar cascades face detection crop faces from images. Run the images through a pre-trained GoogLeNet.

Specifications: Python version 3.7 torch version 1.6 OS: Windows-10-10.0.19041-SP0 Command: CPU: conda install pytorch torchvision torchaudio cpuonly -c pytorch GPU: conda install pytorch torchvision torchaudio cudatoolkit=11.0 -c pytorch

GPU specs used for testing: GPU: Nvidia 2080. NVIDIA CUDA® Cores: 2944 RTX-OPS: 57T Giga Rays/s: 8 Boost Clock (MHz): 1710 Base Clock (MHz): 1515 Memory Speed: 14 Gps Standard Memory Config: 8 GB GDDR6 Memory Interface Width: 256-bit Memory Bandwidth (GB/sec): 448 GB/s