| Jul 03, 2026 |  PorTEXTO is publicly released — a European Portuguese benchmark for visual text extraction. Dataset · Paper. |
| Jul 01, 2026 | AMALIA-VL is publicly released — the first open vision-language model native to European Portuguese. Weights & data · Eval suite · Paper. |
| Nov 10, 2025 |  FineVAU accepted at AAAI 2026 — fine-grained video anomaly understanding with LVLMs. Paper · FineW3 dataset. |
| Mar 20, 2025 | Self-ReS accepted at ICME 2025 — LVLM self-reflection for efficient long-video understanding. Paper. |
| Oct 25, 2024 |  Received the Best Poster Award at RECPAD 2024. |