Welcome to the Depth Estimation repository! This repository houses various implementations of Depth Estimation models tailored for advanced research at the master's level. Below, we present a brief overview of each model along with relevant links to the original papers for further reference.

TinyHitnet is a depth estimation model designed for stereo images, which consist of left and right pairs. This model architecture offers efficient depth estimation while leveraging stereo image inputs.

The Unet architecture, originally proposed for segmentation tasks in the paper U-Net: Convolutional Networks for Biomedical Image Segmentation, has been adapted for monocular depth estimation in this implementation. Unet and its variants provide a robust framework for depth estimation from single images.

CycleGan-Pix2Pix encompasses the implementation of Generative Adversarial Networks (GAN) for monocular depth estimation. This model architecture exploits the power of GANs to generate realistic depth maps from single images.

Stable Diffusion offers an innovative approach to monocular depth estimation utilizing diffusion-based models. The implementation in this directory utilizes stable diffusion models from Hugging Face, tailored specifically for depth estimation tasks.

Each model directory contains the necessary scripts and documentation to train and evaluate the respective depth estimation models. Refer to the README files within each directory for detailed instructions on usage, training, and evaluation.

Thank you for exploring our Depth Estimation repository. For any inquiries or assistance, feel free to reach out. Happy exploring and experimenting with depth estimation techniques!