Embedded Files
Version 1
This dataset includes 72 high-quality images of individuals from the chest up, generated using ComfyUI and enhanced with the flux1-dev model. The images represent diverse body types (fit, middleweight, obese) and emotional expressions (happiness, sadness, neutrality) across genders. Each image was meticulously analyzed and edited to ensure an accurate representation of emotions, with a validation criterion of at least 95% likelihood of the desired emotion. The final dataset features images of four different men and women, each expressing the specified emotions. Additionally, it includes the results of an analysis using Py-Feat, detailing the likelihood of the emotion shown and the specific Action Units (AUs) detected.
Version 2
This dataset includes 96 high-quality images of individuals from the chest up, generated using ComfyUI and enhanced with the flux1-dev model. The images represent diverse body types (fit, middleweight, obese) and emotional expressions (happiness, sadness, anger, and neutrality) across genders. Each image was meticulously analyzed and edited to ensure an accurate representation of emotions, with a validation criterion of at least 95% likelihood of the desired emotion. The final dataset features images of four different men and women, each expressing all four emotions. Additionally, it includes the results of an analysis using Py-Feat, detailing the likelihood of the emotion shown and the specific Action Units (AUs) detected.
Google Sites
Report abuse