In the next work an automatic fringe projection system with a mobile camera has been proposed. A set of experimental images in order to scanning all the set positions in an automated rail are obtained. In each position, four profileometric images shifted pi/2 were taken and use the phase shift of 4 steps algorithm to obtain the wrapped phase. A Radon transform analysis was applied to quantify the object move using the gray levels as scale. A qualitative process to analyze the profilometry of an object with an irregular surface is applied, getting good results.

Generative Adversarial Networks are powerful generative models that are used in different areas and with multiple applications. However, this type of model has a training problem called mode collapse. This problem causes the generator to not learn the complete distribution of the data with which it is trained. To force the network to learn the entire data distribution, MSSGAN is introduced. This model has multiple generators and distributes the training data in multiple subspaces, where each generator is enforced to learn only one of the groups with the help of a classifier. We demonstrate that our model performs better on the FID and Sample Distribution metrics compared to previous models to avoid mode collapse. Experimental results show how each of the generators learns different information and, in turn, generates satisfactory quality samples.

The characterization of emergent collective behavior in chaotic oscillator networks remains a fundamental challenge in nonlinear science. While the temporal evolution of such systems is governed by complex coupling, observing the full phase space is often infeasible. This work investigates the mapping between synchronization manifolds and the underlying structural topology of these networks. Rather than treating neural networks as black-box approximators, we propose a framework in which Long Short-Term Memory (LSTM) units are constrained by synchronization descriptors, such as phase-coherence metrics and Lyapunov exponents. We demonstrate that the model’s internal representations, when initialized with physically-consistent pretraining strategies, mirror the system’s dynamical constraints. This enables the analytical reconstruction of latent functional relationships that are not accessible via standard time-series analysis. Our findings provide new insights into the symmetry-breaking processes and connectivity patterns that govern synchronization stability, bridging the gap between empirical data-driven inference and the formal theory of networked dynamical systems.