Intelligent Systems

The Intelligent Systems group is focused on building solutions for business fields where intelligent ICT is emerging. It is composed of researchers with in-depth expertise in areas such as Signal & Image Processing, Artificial Intelligence and Cognitive Computing: the nuclear areas required to develop systems with the ability to perceive, analyse, learn and ultimately reason about the environments they operate in.

The Intelligent Systems group is divided in four sub-areas, which map our core competences:

 

/ Edge and cloud Computer Vision

We develop edge and cloud computer vision applications not only for image acquisition and processing, but also for identifying the relationship between multiple objects in images or videos. Our team is skilled in the development of portable and affordable image acquisition devices using advanced techniques that culminate in a better understanding of visual contents, which represents a quantum leap in several application areas.

The team is focused on mobile and embedded technology, which facilitate the creation of field-portable, cost-effective imaging and sensing technologies that are approaching laboratory-grade instrument performance. Additionally, we have been developing and optimising advanced algorithms while considering the technological constraints of edge devices, such as small-form IoT, wearables, mobile phones, and embedded processors. 

 

/ Sensor fusion & Embedded intelligence

Information-gathering devices and sensors are now ubiquitous due to the explosive growth of smart devices and the emergence of the Internet of Things (IoT). We at Fraunhofer AICOS use sensor data fusion to integrate multiple data representing the same real world object into a consistent, accurate and useful representation.

Our team came to specialize in higher-levels of data fusion, for instance, through the application of advanced machine learning and statistical processes. By doing this, we maximise the value of data coming from multiple sources, and altogether, we support intelligent system behaviors. By fusing sensors, along with embedded processing and connectivity, we enable context awareness and a plethora of new services for the benefit of society. 

 

/ Cognitive systems and Deep learning 

At Fraunhofer AICOS, we develop cognitive systems inspired in the way the human brain works as well as its capabilities to learn from past experiences to solve highly complex tasks.

We research different information extraction techniques such as text and natural language processing, computer vision and signal processing, to extract information from unstructured sources of information. We combine and integrate multiple sources of information, so that our cognitive systems can understand the meaning and the relation between relevant concepts in any context.

We are also committed to addressing the legal compliance from the initial stages of development, to ensure that dataset bias and unfairness are minimised, and that the transparency of machine-decisions as well as that our solutions have a positive impact in the world.  

 

/ Predictive modelling & Recommendation 

Intelligent systems based on predictive modelling use data to assess certain conditions over time and, based on previous events, model their course and predict future states. At Fraunhofer AICOS, we build predictive modelling and recommendation based on intelligent systems, that are robust and reliable, but also transparent, self-explanatory and easily interpretable for humans.

We focus on computer-aided detection and decision support systems to assist professionals in their routines, for example, early diagnosis and screening, by integrating their knowledge into computer systems.   

Highlighted Projects

 

EyesOnTraps

EyesOnTraps proposes a mobile solution for plague prevention, supporting local temperature registration, automated insect detection in traps and treatment recommendation for detected plagues.

 

SINATRA

SINATRA will allow the combination of the knowledge between the different technical and scientific valences in order to develop an innovative, robust and versatile solution, able to streamline and facilitate industrial preventive and corrective processes.

 

TAMI

TAMI will build high-performing XAI systems for application in medicine, whose workings and outputs can be understood and cross-examined by human users, so that human-friendly explanations of the automated results are also provided. 

 

Further information

 

Competence Articles

 

TSFEL: Time Series Feature Extraction Library

 

Relevant Services

 

Rapid Prototyping

Innovation Studies

Education & Training

Machine Learning Training

 

Relevant Publications

 

Silva, J., Gomes, D., Sousa, I., & Cardoso, J.S. (2020). Automated Development of Custom Fall Detectors: Position, Model and Rate Impact in Performance, 20(10), 5465-5472. DOI: 10.1109/JSEN.2020.2970994. More info

Silva, J., Sousa, I., & Cardoso, J.S. (2020). Fusion of Clinical, Self-Reported, and Multisensor Data for Predicting Falls. In Journal of Biomedical And Health Informatics, 24(1), 50-56. DOI: 10.1109/JBHI.2019.2951230. More info

Folgado D., Barandas M., Matias R., Martins R., Carvalho M. Â., & Gamboa H. (2018). Time Alignment Measurement for Time Series. In Pattern Recognition, 81, 268-279. DOI: 10.1016/j.patcog.2018.04.003. More info

Rodrigues, J., Folgado, D., Belo, D., & Gamboa, H. (2019). SSTS: A syntactic tool for pattern search on time series. In Information Processing and Management, 56(1), 61-76. DOI: 10.1016/j.ipm.2018.09.001. More info

Martins, J., Cardoso, J.S., & Soares, F. (2020). Offline computer-aided diagnosis for Glaucoma detection using fundus images targeted at mobile devices. In Computer Methods and Programs in Biomedicine, 192. DOI: 10.1016/j.cmpb.2020.105341. More info

Andrade, C., Teixeira, L.F., Vasconcelos, M.JM., & Rosado, L. (2020). Deep Learning Models for Segmentation of Mobile-Acquired Dermatological Images. In International Conference on Image Analysis and Recognition, 228-237. DOI: 10.1007/978-3-030-50516-5_20. More info

Costa, J., Pereira, A., & Ferreira, L. (2019). Automatic administration of semantic verbal fluency tests for Portuguese. In International Conference of Experimental Linguistics. DOI: 10.36505/ExLing-2019/10/0015/000377. More info