3 papers - avg viability 6.7
A novel unsupervised video anomaly detection method that uses an entropy-guided autoencoder to achieve superior performance by minimizing latent space entropy for clearer anomaly identification.
TIMID is a novel architecture for detecting time-dependent mistakes in robot execution videos using weak supervision.
Develop a multimodal large language model-based tool for video anomaly detection in surveillance with class-specific prompting.