Back To Projects

Predicting emotion ratings from color statistics of images

Claire M.

We utilized this data to see how visual elements affect our everyday lives by engineering features of images and running that through a variety of neural networks.


Which features in art affect emotional reactions? In order to answer this question, we took a variety of abstract images and included people’s emotional ratings for each image, for eight different emotions. We utilized this data to see how visual elements affect our everyday lives by engineering features of images and running that through a variety of neural networks. Out of all the networks we ran our data through, the linear regression model performed the best with a mean average error of around 1.4. All the other neural networks, including a convolutional neural network, had a mean average error of at least 1.7. The full-size linear regression model with all engineered features worked best in predicting the emotional values associated with each image. That is, all specific features were significant and we did not find meaningful differences among the models for different emotion ratings.

Explore More!

Claire M.
Clayton Greenberg
Teaching Professor from UPenn

Related Projects

The Impacts of Child-Mentor Relationships on Child Mental Health

In this paper, we use a dataset from the Substance Abuse and Mental Health Data Archive to analyze the impacts of child-mentor relationships on child mental health. Previous studies have come to similar conclusions: positive parent-student and teacher-student relationships often lead to signs of positive mental health in children. To expand upon these past findings, we used different measures of mental health and relationship qualities to create a predictive model. We conducted classification and coefficient weight analyses to see how strong of an impact different variables representing indicators of healthy relationships had on different aspects of child mental health. This information was used to create predictions of future cases. Like previous studies, we found that there was a general pattern that showed that good child-mentor relationships, defined specifically by the frequency of praise and fights, had an overall positive impact on child mental health, specifically when it came to symptoms of depression. Furthermore, parents seemed to have stronger impacts on child mental health than teachers did. Limitations include possibly biased respondents who may not be representative of the greater population, as well as the specificity of the variables that were chosen. Going forward, further steps to analyze different datasets and deepen the scope of the research will be helpful in finding patterns and developing more detailed conclusions.
Samantha K.
Mentored by Akshay Jagadeesh
Training a Machine Learning Model to Recognize Signs of Depression

Many people have not had a professional diagnosis, and could have depression without being aware of it. To attempt to answer this question, I used a person’s daily habits and how they change, and attempt to predict if they will experience a depressive episode in the future.
Bwohan W.
Mentored by Barbie Duckworth
How Does One’s Background Determine Their Mental State?

Mental health issues have become very prevalent in recent times, and although significant progress has been made in terms of treatment in the form of counseling, medication, and other methods, in order to truly find out the root cause of many mental health issues, a correlation has to be drawn between one’s mental state and another factor, such as family history, age, work environment, etc. It is also beneficial to correlate one’s mental state with a multitude of factors to see the compilation.
Sohum T.
Mentored by Akshay Jagadeesh