MIVIA Lab

The repository contains the benchmark, the dataset, the annotations and the code to reproduce the results reported in the paper: V. Carletti, A. Greco, M. Marseglia, M. Vento, "Context Matters: Towards Adaptable Open Vocabulary Video Event Detection in the Wild", European Conference on Computer Vision Workshops 2026.

This is the codebase repository for "Scaling Subgraph Isomorphism: Exploiting Massive GPU Parallelism with VF-GPU"

Codebase for paper "ADAPT-MAGNET: A Multimodal Multitask Architecture for Gender and Emotion Recognition"

This repository contains the dataset described in the following paper: Greco A., Ricciardi A., Sansone C., Vento B., "Benchmarking illegal waste dumping detection: a public video dataset and a reference baseline", Image and Vision Computing.

This repository will contain the dataset and the code related to the following paper: Greco A., Marseglia M., Vento M., "Online fully supervised video anomaly detection", Submitted to Neural Computing and Applications. The dataset and code will be made available after the acceptance of the paper.

Official repository supporting OV-CAVED, a context-aware open-vocabulary video event detection framework. It provides the automatic LLM-based annotation tool and links to the OV-CAVED UCF-Crime benchmark released on Zenodo.

This repository will contain the code related to the following paper:

This repository contains a reorganization of the RSCD dataset (https://thu-rsxd.com/rscd/) on a per-acquisition-day basis to assess the generalization capabilities of the methods. We share the CSV files containing the training, validation and test samples with the scientific community.