Depth Map Super-Resolution via Cascaded Transformers Guidance

Depth information captured by affordable depth sensors is characterized low spatial resolution, which limits potential applications. Several methods have recently been proposed for guided super-resolution of maps using convolutional neural networks to overcome this limitation. In a scheme, high-resolution are inferred from low-resolution ones with the additional guidance corresponding intensity image. However, these still prone texture copying issues due improper We propose multi-scale residual deep network map super-resolution. A cascaded transformer module incorporates structural image into upsampling process. The achieves linear complexity in making it applicable images. Extensive experiments demonstrate that method outperforms state-of-the-art techniques