Affective Computing and Applications of Image Emotion Perceptions
Authors: Sicheng Zhao, Hongxun Yao
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The research of my Ph D thesis focuses on image emotion computing (IEC), which aims to predict the emotion perceptions of given images. We demonstrate that the principles-of-art based emotion features (PAEF) can model emotions better and are more interpretable to humans. To further demonstrate this observation, we set up a large-scale dataset, named Image-Emotion-Social-Net dataset, with over 1 million images downloaded from Flickr. The EM algorithm is used to estimate the parameters of GMM. In experiment, the EM algorithm is converged in 6.28 steps on average. |
| Researcher Affiliation | Academia | Sicheng Zhao, Hongxun Yao School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China. zsc@hit.edu.cn, h.yao@hit.edu.cn |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not include an explicit statement about releasing its source code, nor does it provide a link to a code repository for the described methodology. |
| Open Datasets | Yes | The images in Abstract dataset (Machajdik and Hanbury 2010) were labeled by 14 people on average. |
| Dataset Splits | No | The paper describes the creation and size of datasets ('Image-Emotion-Social-Net dataset', '1,434,080 emotion labels on 1,012,901 images'), but it does not provide specific training, validation, or test split percentages or counts, nor does it cite predefined splits for reproducibility. |
| Hardware Specification | No | The paper does not specify any hardware details (e.g., GPU models, CPU types, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions techniques and algorithms like 'EM algorithm' and 'GMM', but it does not list any specific software dependencies or libraries with version numbers (e.g., Python 3.x, PyTorch 1.x) that would be needed for reproducibility. |
| Experiment Setup | Yes | Specifically, the initializations are obtained by firstly partitioning the VA labels into two clusters based on whether valence is greater than 5 and then computing the mean vector μl and covariance matrix Σl of each cluster. The mixing coefficients are set as the proportions of related VA labels in each cluster to the total labels. In experiment, the EM algorithm is converged in 6.28 steps on average. |