Training Deep Networks for Facial Expression Recognition with Crowd-Sourced Label Distribution

Barsoum, Emad; Cha, Zhang; Ferrer, Cristian Canton; Zhang, Zhengyou

doi:10.48550/arxiv.1608.01041

Cited by 6 publications

(7 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results show a similar behavior as the one previously shown for Affectnet. The main difference is a higher general accuracy (top accuracy 83.2%), consistent with the results of [Barsoum et al, 2016] for this dataset.…”

Section: A3 Model Bias Resultssupporting

confidence: 87%

“…The final images have 425 by 425 pixels of resolution. Appendix A repeats the same analysis for a second dataset, FER+ [Barsoum et al, 2016].…”

Section: Datasetmentioning

confidence: 97%

“…For completeness, we repeat the analysis on a second dataset, FER+ [Barsoum et al, 2016]. FER+ is based on FER2013 [Goodfellow et al, 2013], a large ITW dataset composed of 32,298 color images.…”

Section: A Fer+mentioning

confidence: 99%

See 2 more Smart Citations

Assessing Demographic Bias Transfer from Dataset to Model: A Case Study in Facial Expression Recognition

Dominguez-Catena¹,

Paternain²,

Galar³

2022

Preprint

View full text Add to dashboard Cite

The increasing amount of applications of Artificial Intelligence (AI) has led researchers to study the social impact of these technologies and evaluate their fairness. Unfortunately, current fairness metrics are hard to apply in multi-class multidemographic classification problems, such as Facial Expression Recognition (FER). We propose a new set of metrics to approach these problems. Of the three metrics proposed, two focus on the representational and stereotypical bias of the dataset, and the third one on the residual bias of the trained model. These metrics combined can potentially be used to study and compare diverse bias mitigation methods. We demonstrate the usefulness of the metrics by applying them to a FER problem based on the popular Affectnet dataset. Like many other datasets for FER, Affectnet is a large Internetsourced dataset with 291,651 labeled images. Obtaining images from the Internet raises some concerns over the fairness of any system trained on this data and its ability to generalize properly to diverse populations. We first analyze the dataset and some variants, finding substantial racial bias and gender stereotypes. We then extract several subsets with different demographic properties and train a model on each one, observing the amount of residual bias in the different setups. We also provide a second analysis on a different dataset, FER+.

show abstract

Section: A3 Model Bias Resultssupporting

confidence: 87%

“…The final images have 425 by 425 pixels of resolution. Appendix A repeats the same analysis for a second dataset, FER+ [Barsoum et al, 2016].…”

Section: Datasetmentioning

confidence: 97%

See 1 more Smart Citation

Assessing Demographic Bias Transfer from Dataset to Model: A Case Study in Facial Expression Recognition

Dominguez-Catena¹,

Paternain²,

Galar³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…To allow conversion of a model from .onnx format to .wasm format, the ONNX variant should adhere to the following constraints: ONNX version 1.7.0, opset 12, file-format 7 5 . The Scailable python package 6 provides a validation function to aid in the conversion process. For the sake of replicability, all subjects are already pre-trained and we provide the same inputs to each of them.…”

Section: Subjects Selectionmentioning

confidence: 99%

“…Even though assessing the accuracy of the models is out of scope of this study, the interested reader can inspect their output in our replication package. MNIST [20] Image Classification MNIST 70k/30k Emotion [6] Image Classification Emotion FER 4k/1.8k CIFAR10 [18] Image Classification CIFAR-10 4k/2k YOLOv4 [7] Object Detection COCO 700/300…”

Section: Subjects Selectionmentioning

confidence: 99%

An empirical study on the Performance and Energy Consumption of AI Containerization Strategies for Computer-Vision Tasks on the Edge

Hampau

Kaptein

Emden³

et al. 2022

The International Conference on Evaluation and Assessment in Software Engineering 2022

View full text Add to dashboard Cite

Context. The rise of use cases of AI catered towards the Edge, where devices have limited computation power and storage capabilities, motivates the need for better understating of how AI performs and consumes energy. Goal. The aim of this paper is to empirically assess the impact of three different AI containerization strategies on the energy consumption, execution time, CPU, and memory usage for computer-vision tasks on the Edge. Method. In this paper we conduct an experiment with the used containerization strategy as main factor, with three treatments: ONNX Runtime, WebAssembly, and Docker. The subjects of the experiment are four widely-used computer-vision algorithms. We then orchestrate a series of runs where we deploy the four subjects on different generations of Raspberry Pi devices, with different hardware capabilities. A total of 120 runs (per device) are recorded to gather data on energy, execution time, CPU, and memory. Results. We found a statistically significant difference between the three containerization strategies on all dependent variables. Specifically, WebAssembly proves to be a valuable alternative for devices with reduced disk space and computation power. Conclusions. For computer-vision tasks with limited disk space and RAM memory requirements, developers should prefer WebAssembly for deployment. The (non-dockerized) ONNX Runtime resulted to be the best choice in terms of energy consumption and execution time.

show abstract

Saliency map and deep learning based efficient facial emotion recognition technique for facial images

Kumari

Bhatia

2023

Multimed Tools Appl

View full text Add to dashboard Cite

Training Deep Networks for Facial Expression Recognition with Crowd-Sourced Label Distribution

Cited by 6 publications

References 23 publications

Assessing Demographic Bias Transfer from Dataset to Model: A Case Study in Facial Expression Recognition

Assessing Demographic Bias Transfer from Dataset to Model: A Case Study in Facial Expression Recognition

An empirical study on the Performance and Energy Consumption of AI Containerization Strategies for Computer-Vision Tasks on the Edge

Saliency map and deep learning based efficient facial emotion recognition technique for facial images

Contact Info

Product

Resources

About