Papers accepted to WACV2025!
We are thrilled to announce that two of our research papers have been accepted to the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2025)! WACV is an international event that showcases cutting-edge research in computer vision, offering a unique platform for students, academics, and industry researchers. We are honored to present our work among some of the brightest minds in the field between March 1st and 3rd, 2025. More info can be found under the official page for WACV 2025. Accepted papers from our group and their short description are as follows:
Quercia, Alessio, Erenus Yildiz, Zhuo Cao, Kai Krajsek, Abigail Morrison, Ira Assent, and Hanno Scharr. "Enhancing Monocular Depth Estimation with Multi-Source Auxiliary Tasks." In Winter Conference on Applications of Computer Vision, WACV. 2025.
Monocular depth estimation (MDE) is a challenging task in computer vision, often hindered by the cost and scarcity of high-quality labeled datasets. We tackle this challenge using auxiliary datasets from related vision tasks for an alternating training scheme with a shared decoder built on top of a pre-trained vision foundation model, while giving a higher weight to MDE. Through extensive experiments we demonstrate the benefits of incorporating various in-domain auxiliary datasets and tasks to improve MDE quality on average by ~11%, and that auxiliary tasks have different impacts. Remarkably, our study reveals that using semantic segmentation datasets as Multi-Label Dense Classification (MLDC) often results in additional quality gains. Lastly, our method significantly improves the data efficiency for the considered MDE datasets, enhancing their quality while reducing their size by at least 80%.
Arya Bangun, Zhuo Cao, Alessio Quercia, Hanno Scharr, and Elisabeth Pfaehler. "MRI Reconstruction with Regularized 3D Diffusion Model (R3DM)." In Winter Conference on Applications of Computer Vision, WACV. 2025.
Magnetic Resonance Imaging (MRI) is widely used to visualize internal structures but requires faster 3D reconstruction algorithms for under-sampled data. We propose a 3D MRI reconstruction method combining a regularized 3D diffusion model with optimization techniques to address this need. By using diffusion-based priors, our approach enhances image quality, reduces noise, and improves fidelity. Comprehensive experiments on clinical and plant science MRI datasets validate its performance across various undersampling patterns and data distributions.
Stay tuned for more details about our presentations at WACV 2025!