Researchers are exploring various methods for crack segmentation and detection, including the use of enhanced models such as Mamba and U-Net architectures. Different approaches, such as the use of vision transformers and gradient transformer self-attention, are being investigated for their effectiveness in detecting cracks in surfaces like concrete bridges and roads. The development of lightweight models, such as those based on improved MobileNetV3 and VM-UNet, is also being researched for efficient crack recognition and segmentation.