This paper presents an Automatic Face Analysis (AFA) system designed to recognize fine-grained changes in facial expressions using the Facial Action Coding System (FACS). The system focuses on both permanent and transient facial features, including brows, eyes, mouth, and furrows, in nearly frontal-view face image sequences. It employs multistate face and facial component models to track and model various facial features, extracting detailed parametric descriptions during tracking. These parameters are then fed into neural network-based classifiers to recognize action units (AUs) of the upper and lower faces, either alone or in combinations. The AFA system achieves average recognition rates of 96.4% for upper face AUs and 96.7% for lower face AUs, with high generalizability across independent databases. The system's performance is compared with other AU recognition systems, showing superior or comparable results. The AFA system represents a significant advancement in the field of automatic facial expression analysis, providing a more comprehensive and accurate method for recognizing subtle changes in facial expressions.This paper presents an Automatic Face Analysis (AFA) system designed to recognize fine-grained changes in facial expressions using the Facial Action Coding System (FACS). The system focuses on both permanent and transient facial features, including brows, eyes, mouth, and furrows, in nearly frontal-view face image sequences. It employs multistate face and facial component models to track and model various facial features, extracting detailed parametric descriptions during tracking. These parameters are then fed into neural network-based classifiers to recognize action units (AUs) of the upper and lower faces, either alone or in combinations. The AFA system achieves average recognition rates of 96.4% for upper face AUs and 96.7% for lower face AUs, with high generalizability across independent databases. The system's performance is compared with other AU recognition systems, showing superior or comparable results. The AFA system represents a significant advancement in the field of automatic facial expression analysis, providing a more comprehensive and accurate method for recognizing subtle changes in facial expressions.