TAEHV is a Tiny AutoEncoder for Hunyuan Video (and other similar video models). TAEHV can encode and decode latents into videos more cheaply (in time & memory) than the full-size video VAEs, at the ...
Abstract: This article presents a deep autoencoder-based methodology for unsupervised anomaly detection in centrifugal pumps under limited failure data conditions, focusing on real-world applications ...
Abstract: There is a rising necessity to enhance advanced methods of image processing through the noise reduction using improved neural network topologies. Autoencoder networks, untied types of neural ...
A comprehensive Edge AI and IoT anomaly detection system designed for research and educational purposes. This project demonstrates real-time anomaly detection using autoencoder-based models optimized ...