DataTrainX is a prototype of Facial Emotion Recognition (FER) which main objective is to focus on the analysis of any learner behavior for the purpose of neuropedagogy.
In learning, the logic of restitution which still prevails over the logic of comprehension would be at the origin of many failures of the learner. To understand themselves, the world and others, all learners produce and implement metacognitive resources. We can notice that this calls upon many disciplines, such as psychology, pedagogy or neurosciences. DataTrainX is a prototype of Facial Emotion Recognition (FER) which main objective is to focus on the analysis of any learner behavior for the purpose of neuropedagogy. In order to achieve this, we will use deep-Learning algorithms and more particularly convolutional neural networks (CNN) in recognition of basic emotions defined by the psychologist P. Ekman. The idea is to focus on the approach that each individual favors to apprehend the act of learning. In the DataTrainx project the user will be recorded on video and a comparison will be made between the typology found by the Kolb questionnaire and the emotions he has expressed. The objective is to find a correlation between the learning typology and the user's emotion. The purpose of this project is to focus concretely on the IT functional means to be implemented for this type of application and being to demonstrate how to set up the functional, software and hardware architecture to arrive at an application result of FER.
Key words : CNN, REF, Deep learning, Tensorflow, Dataset, Convolution, NodeJS, learning, neurosciences, neuropedagogy, Kolb, video.
For more information, check out our manuscript:
DataTrainX: Researching the cognitive singularities of a learner, based on deep-learning. Davy Blavette 2022
For a live demo, visit: https://datatrainx.akairnet.fr/
Clone or download this repository:
git clone git@github.com:davy-blavette/DatatrainX.git
Install the dependencies:
npm installThen run DataTrainX:
npm run devTo see how we trained the CNN, visit the directory ./dataset/fer2013.
DataTrainX was created by Davy Blavette and thanks contribution CNN Explainer Jay Wang Face-API Vincent Mühler
which was the result of a research collaboration between CNAM IRIT Toulouse
The software is available under the MIT License.
If you have any questions, feel free to open an issue.