ICCT Europe 2025

On April 28th, Javier Usón presented the work “Is Real-Time Deep Learning-Based Monocular Depth Estimation Accurate for Multi-Camera Setups?” at ICCT Europe in Faro, in the session "Network Security and Virtual Reality", chaired by Julián Cabrera.

The talk focused on evaluating state-of-the-art monocular depth estimation techniques within the context of real-time volumetric video capture. Instead of analyzing the raw volumetric data, the performance was assessed through the quality of the final rendered synthetic views using the FVV Live system.

While the results were encouraging, several challenges remain, particularly artifacts at scene boundaries and double-image effects due to camera misalignment. Still, this opens exciting directions for refining depth estimation in real-time multi-camera environments.