Please enable JavaScript.
Coggle requires JavaScript to display documents.
The Potential High Priority Care based on Pain Facial Expression Detection…
The Potential High Priority Care based on Pain Facial Expression Detection with Chest Pain Patients
Dataset
2 classes: 1450 Pain and 1500 No Pain
Exclude: dark skin or inadequate light
Materials and Methods
Method YOLO
YOLOv5 (nano, small, medium, large, x)
YOLOv5n6,YOLOv5s6, YOLOv5m6, YOLOv5l6,YOLOv5x6+TTA
YOLOv4
YOLOv6
YOLOv7
Material: Camera
Environment: Windows10 and Linux Ubuntu 18.04
Python??, KERAS, TensorFlow
Labeling Roboflow
Key Point: eyebrows and eyes
Data augmentation
70% data training: 15% data validation: 15% data testing
Data Primer: 1000 videos
Performance
YOLOv5l
high accuracy, FLOP largerst, training time long