Urban ecosystems are suitable for the introduction and spread of alien bird species, and early detection of their establishment and expansion is crucial to reduce potential negative impacts. In this context, the use of opportunistic citizen science data can have considerable advantages in relation to conventional scientific approaches. We gathered records of parakeets and parrots (Psittaciformes) and mynas and ...
Wildlife roadkill databases often compile valuable information regarding species, timing, and locations of roadkill incidents, with some databases also including details such as gender and age classes. As part of the “LIFE LINES - Linear Infrastructure Networks with Ecological Solutions” project, we aggregated roadkill data from various sources, including road operators, traffic police, and academic institution...
The need for reversing the negative impacts of Linear Infrastructure Networks on the environment is increasingly acknowledged as a key to achieving national and international biodiversity commitments. LIFE-LINES project (LIFE14 NAT/PT/001081) employed and essayed a large and diverse number of interventions to reconcile biodiversity conservation and the presence of roads and powerlines. This was achieved through...
This work investigates the use of LLMs to enhance automation in software testing, with a particular focus on generating high-quality, context-aware test scripts from natural language descriptions, while addressing both text-to-code and text+code-to-code generation tasks. The Codestral Mamba model was fine-tuned by proposing a way to integrate LoRA matrices into its architecture, enabling efficient domain-specif...
This study presents a systematic exploration of strategies for pretraining generative Large Language Models (LLMs) within the Galician-Portuguese diasystem, by focusing on two underrepresented varieties of this diasystem, namely European Portuguese and Galician. We investigate the impact of combining versus separating linguistic varieties during continued pretraining, the trade-offs between large-scale noisy da...
The biodiversity impacts of agricultural deforestation vary widely across regions. Previous efforts to explain this variation have focused exclusively on the landscape features and management regimes of agricultural systems, neglecting the potentially critical role of ecological filtering in shaping deforestation tolerance of extant species assemblages at large geographical scales via selection for functional t...
The automatic detection of suspicious abandoned objects has become a priority in video surveillance in the last years. Terror- ist attacks, improperly parked vehicles, abandoned drug packages and many other events, endorse the interest in automating this task. It is challenge to detect such objects due to many issues present in public spaces for video-sequence process, like occlusions, illumination changes, cro...
Recently, the Convention on Biological Diversity Conference of the Parties (COP 15) adopted the “post-2020 global biodiversity framework” where inverting biodiversity decline and restoring ecological connectivity are major flagships. The need for reversing the negative impacts of Linear Infrastructure Networks in the environment is increasingly acknowledged as a key to achieving those biodiversity commitments. ...
As poeiras provenientes do deserto (Norte de África) assolam com alguma frequência o território português e constituem um dos eventos naturais com maior predominância. A observação deste evento traduz-se pela poeira que se deposita nas superfícies (carros, casas, ruas) e, também, pelos efeitos que provoca na radiação solar. O registo fotográfico permite constatar o efeito ótico gerado, que é comprovado pela med...
We have developed an ASR system for European Portuguese implement ing the QuartzNet [3] architecture with the NeMo [4] framework. Two approaches were used in this work: from scratch and using transfer learning. The experiments were data-driven focused instead of algorithm finetuning. Experiments confirm that models developed using transfer learning have shown better results (WER=0.0513) than developing models f...