dissertation
:mortar_board: :scroll: This repository holds my final year and dissertation project during my time at the University of Lincoln titled 'Deep Learning for Emotion Recognition in Cartoons'.
Science Score: 44.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
✓CITATION.cff file
Found CITATION.cff file -
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
○DOI references
-
○Academic publication links
-
○Committers with academic emails
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (13.5%) to scientific vocabulary
Keywords
Repository
:mortar_board: :scroll: This repository holds my final year and dissertation project during my time at the University of Lincoln titled 'Deep Learning for Emotion Recognition in Cartoons'.
Basic Info
- Host: GitHub
- Owner: hako
- License: bsd-2-clause
- Language: TeX
- Default Branch: master
- Homepage: https://hako.github.io/dissertation
- Size: 45.5 MB
Statistics
- Stars: 25
- Watchers: 3
- Forks: 19
- Open Issues: 0
- Releases: 0
Topics
Metadata Files
README.md
dissertation
This repository holds my final year project during my time at the University of Lincoln titled 'Deep Learning for Emotion Recognition in Cartoons'.
Abstract
Emotion Recognition is a field that computers are getting very good at identifying; whether it's through images, video or audio. Emotion Recognition has shown promising improvements when combined with classifiers and Deep Neural Networks showing a validation rate as high as 59% and a recognition rate of 56%.
The focus of this dissertation will be on facial based emotion recognition. This consists of detecting facial expressions in images and videos. While the majority of research uses human faces in an attempt to recognise basic emotions, there has been little research on whether the same deep learning techniques can be applied to faces in cartoons.
The system implemented in this paper, aims to classify at most three emotions (happiness, anger and surprise) of the 6 basic emotions proposed by psychologists Ekman and Friesen, with an accuracy of 80% for the 3 emotions. Showing promise of applications of deep learning and cartoons. This project is an attempt to examine if emotions in cartoons can be detected in the same way that human faces can.
Dataset
The dataset used in this dissertation is a collection of 4,800 Tom & Jerry face images.
Tom & Jerry Image Dataset (15MB)
Requirements
- Python 2.7
- OpenCV 3.2+
- TensorFlow 1.1+ CPU/GPU (GPU Recommended)
- Jupyter Notebook (Optional)
- Linux:
sudo apt-get install python-dev python-tk
Install
git clone https://github.com/hako/dissertation
cd dissertation
sudo pip install -r requirements.txt
Usage
Download the above dataset, the folder must be named datasets. Below you can get started with the tools below.
Training / Classification / Visualisation
If you just want to train/classify or visualise the output of the network, use this tool:
``` training: (and show summary or results) usage: train.py -t [-v|-s]
classification: usage: train.py -c image.jpg
visualisation: usage: train.py -V ```
Segmentation
Included in this repo are two Haar cascade files trained to detect Tom & Jerry faces. Note that if you choose this tool, you have to obtain the Tom & Jerry videos yourself.
If you want to segment the Tom & Jerry videos into images, use this tool:
python segmentation.py
Notebook
If you're the type that likes interactivity and experimentation, both the segmentation.py and the train.py files have their own Jupyter Notebooks in the notebooks/ folder. If you're using this, make sure the video and image datasets are in the folder.
Special Thanks
I would like to thank the following:
- Professor Stefanos Kollias
- My family and friends
- The University of Lincoln Library
Citation
Hill, J.W., (2017). Deep Learning for Emotion Recognition in Cartoons
bibtex
bibtex
@mastersthesis{hill17,
author = {John Wesley Hill},
title = {Deep Learning for Emotion Recognition in Cartoons},
school = {University of Lincoln},
year = {2017},
document_type = {Bachelor's Thesis},
type = {Bachelor Thesis},
}
Notes
Experiment Environment:
OS: Ubuntu GNU/Linux 16.04 LTS (x64)
GPU: Nvidia GeForce GTX 970
Tom & Jerry © Warner Bros. Entertainment, Inc
License
BSD
Owner
- Name: Wesley Hill
- Login: hako
- Kind: user
- Location: london
- Website: https://wesleyhill.co.uk
- Repositories: 25
- Profile: https://github.com/hako
software eng + animator (hobbyist). @mytopschools @fullcycleapp
Citation (CITATION.cff)
cff-version: 1.2.0 message: If you use this software, please cite it as below. authors: - family-names: Hill given-names: John Wesley affiliation: The University of Lincoln, UK title: Deep Learning for Emotion Recognition in Cartoons version: 1.0.0 date-released: 2017-06-05 url: https://hako.github.io/dissertation repository-code: https://github.com/hako/dissertation license: BSD-2-Clause
GitHub Events
Total
Last Year
Committers
Last synced: about 1 year ago
Top Committers
| Name | Commits | |
|---|---|---|
| Wesley Hill | w****y@h****k | 2 |
Committer Domains (Top 20 + Academic)
Issues and Pull Requests
Last synced: 10 months ago
All Time
- Total issues: 0
- Total pull requests: 1
- Average time to close issues: N/A
- Average time to close pull requests: 3 minutes
- Total issue authors: 0
- Total pull request authors: 1
- Average comments per issue: 0
- Average comments per pull request: 0.0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Issue authors: 0
- Pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Top Authors
Issue Authors
Pull Request Authors
- krutik2966 (1)
Top Labels
Issue Labels
Pull Request Labels
Dependencies
- Keras ==2.0.4
- Pillow ==4.1.1
- jupyter ==1.0.0
- keras-vis ==0.3
- matplotlib ==2.0.2
- numpy ==1.12.1
- opencv-python ==3.2.0.7
- tensorflow ==1.1.0