The field of computer vision is witnessing significant advancements in person re-identification and intention prediction, driven by the development of innovative architectures and techniques. A key direction in this field is the integration of Vision Transformers (ViTs) with other models, such as ConvNeXt, to leverage their complementary strengths and improve performance in complex scenarios like occlusion and viewpoint distortion. Another area of focus is the development of occlusion-aware models that can effectively handle incomplete observations and predict pedestrian intentions. Noteworthy papers in this area include: The Sh-ViT model, which achieves state-of-the-art performance in occluded person re-identification by introducing a Shuffle module and scenario-adapted augmentation. The ConvNeXt-ViT hybrid architecture, which demonstrates superior performance in facial age estimation by combining the strengths of CNNs and ViTs. The Occlusion-Aware Diffusion Model, which reconstructs occluded motion patterns to guide future intention prediction and achieves robust performance under various occlusion scenarios.