Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Due to their impact on emotional health, mental functioning, productivity and quality of life, depression and mental fatigue have become public health problems. Early diagnosis and early intervention are important to the successful treatment of these conditions. Thanks to the recent developments in Artificial Intelligence (AI), Computer Vision (CV), Affective Computing and Explainable Artificial Intelligence (XAI), automatic systems have been developed that are able to analyze behavioral signs associated with psychological conditions. This review provides an overview and detailed analysis of recent research on facial expression recognition, facial micro-expression analysis, monitoring of eye dynamics, depression detection, mental fatigue assessment, multimodal learning and explainable AI. The literature is systematically organized according to application area, data sets, methodology and performance. Comparative, dataset, technology and technique-wise analysis is carried out to find out the trend, strengths and limitations of the existing research. The results show that deep learning-based architectures like CNN, LSTM, CNN-LSTM hybrid, GNN, and transformer models can achieve a significant boost in behavioral analysis performance compared to single-modality models, and multimodal models tend to perform better than their single-modality counterparts. The use of techniques such as SHAP, Grad-CAM, and Integrated Gradients further enhances the transparency and trustworthiness of AI models. Previous research tends to focus on the detection of depression or fatigue separately, but not together with say, facial micro-expressions, eye dynamics, and explainability. Considering the deficiencies in the literature, an Explainable Temporal Vision Framework is proposed to simultaneously detect silent depression and mental fatigue using facial micro-expressions and eye behavior. The review highlights new challenges, future research directions and possible avenues to develop reliable, scalable and clinically viable mental health monitoring systems.