Infant FLACC Pain Level Video Dataset (IFPaLVD)
This dataset was used in a research for detecting cry and pain level in an infant video using Deep Learning Autoencoder for dimentionality reduction, and Convolutional Neural Networks (CNN) as classifier. This dataset was processed by Natanael Simogiarto, with Dr. Yosi Kristian as the Advisor. This dataset provides three pain level target labels, which are no pain, low/moderate pain, and severe pain. For the crying labels, this dataset provide crying and no crying labels.
The video includes both infant facial expression and infant voice. In original, we have 56 videos with various duration. Later on, these videos were divided into smaller segments, with an average of ten seconds duration for each video. As a result, we achieve 253 videos in totals. However, we have an imbalance amount for each classes. For more details, the label distribution for this dataset can be seen at the below
|Pain Level||Severe Pain||80|