Skip to content

We investigate and create solutions adopting the last breakthrough results and technologies from Academia and companies. 

We are proud of our long experience in the field of Biometrics, specifically in tasks such as fingerprint recognition and face morphing detection attacks. 

We are specialized in image analysis and in developing computer vision-based solutions, ranging from Video Surveillance to Machine and Robot vision.

The Mi@BioLab is a research laboratory at University of Bologna born from the experience matured at the Biometric System Laboratory (that’s why we refer as MI@Biolab). It is located at the Department of Computer Science and Engineering (DISI), Cesena (Italy).

MI@Biolab is specialized in researching and developing solutions in the field of Machine Learning, Computer Vision and Biometrics. Collaborations with industrial partners ensure that the research activities are linked to real applications.

Latest News

CAR GAZE DATASET
CAR GAZE DATASETOctober 5, 2026By Fabio Notaro CARGAZE: AUTOMOTIVE EYE-TRACKING AND SEMANTIC SALIENCY DATASET Download link: COMING SOON 1. OVERVIEW CarGaze is a comprehensive, multi-modal eye-tracking and semantic visual saliency dataset tailored for the automotive domain. It provides high-frequency gaze coordinates, visual attention heatmaps, individual vehicle component semantic masks (both ground truth and predicted), demographic survey data and high-resolution vehicle stimuli. Key Dataset Features Participants: 55 subjects tested in free-viewing conditions (10 seconds per stimulus) Eye-Tracker Hardware: Tobii 4C eye-tracker calibrated at 90Hz Vehicles: 12 representative vehicles balanced across 3 industrial segments (Economical, Electric, Luxury) Viewpoints: 3 canonical perspectives per vehicle (Front, Lateral, Rear) = 36 primary visual stimuli (+ 3 Black & White stimuli for Ferrari Testarossa) Total Validated Participant Trials: 2145 individual trials (55 subjects x 39 trials) Semantic Components: fine-grained vehicle component segmentation masks without composite aggregated images. 2. DATASET STRUCTURE CarGaze/├── README.md                                    # Documentation of the dataset├── metadata/                                       # Subject demographics and perceptual survey evaluations│   └── DesignPerceptionSurvey.csv     # 55 responses: age, gender, automotive/design interest, aesthetic scores├── images/                                          # Original car stimuli (36 PNG images)│   ├── ECONOMICAL/                # Dacia Sandero 2021, Ford Puma 2020, Peugeot 208 2020, VW Golf 2024│   ├── ELECTRIC/                        # Lynk & Co 03 2019, Polestar 2 2020, Polestar 4 2024, Tesla Model 3 2024│   └── LUXURY/                                  # Ferrari 849 Testarossa 2026, Ferrari Amalfi 2026, Lambo Revuelto 2024, Lambo Urus 2019├── semantic_masks/                          # Binary semantic masks of individual vehicle components│   ├── ground_truth/                         # Ground Truth reference masks for all vehicle parts│   │   ├── ECONOMICAL/                  │   │   ├── ELECTRIC/│   │   └── LUXURY/│   └── predicted/                             # Semantic masks generated by our automated segmentation method│       ├── ECONOMICAL/                │       ├── ELECTRIC/│       └── LUXURY/└── participants/                              # Individual eye-tracking results for each subject (1 to 55)    ├── 1/    │   ├── ECONOMICAL/    │   │   └── DACIA/SANDERO 2021/    │   │       ├── results_front/    # Trial gaze files: eyetracking_data.txt, heatmap, scanpath, point cloud    │   │       ├── results_lateral/    │   │       └── results_rear/    │   ├── ELECTRIC/    │   └── LUXURY/                             ├── …    └── 55/ Folder Content Details metadata/: contains demographic and survey information (DesignPerceptionSurvey.csv). For all 55 participants, this records age, gender, self-assessed interest level in automotive and design (scale 0-4) and explicit aesthetic appreciation ratings for specific car components (lights, wheels, logo, grille, body) images/: contains the 36 images (front.png, lateral.png, rear.png) for all 12 vehicles, categorized under ECONOMICAL, ELECTRIC, and LUXURY semantic_masks/ground_truth/: contains reference ground-truth binary masks for each specific car component semantic_masks/predicted/: contains the predicted component binary masks produced by our deep learning segmentation method participants/: contains 55 subdirectories, one per participant 3. SEMANTIC COMPONENTS CATALOG The semantic masks segment individual automotive components into binary masks (0 = background, 255 = component). FRONT VIEW car_brand_badge_emblem car_front_grille car_front_windshield_glass car_hood car_left_front_tire car_right_front_tire car_left_headlight car_right_headlight car_left_lower_bumper_light car_right_lower_bumper_light car_left_side_mirror car_right_side_mirror car LATERAL VIEW front_bumper rear_bumper front_door rear_door front_door_handle rear_door_handle front_door_window rear_door_window front_wheel rear_wheel fuel_door headlight taillight.png rearview_mirror car REAR VIEW logo rear_windshield rear_bumper left_taillight right_taillight left_wheel right_wheel left_mirror right_mirror exhaust_tips car PARTICIPANT DATA FORMAT In each participant trial folder (results_front/, results_lateral/, results_rear/): eyetracking_data.txt: raw tabular gaze samples acquired at 90Hz containing timestamps, normalized screen coordinates (x, y) heatmap.png: continuous visual density saliency map generated via Gaussian kernel convolution over participant fixations scanpath.png: sequential visualization of fixation points connected by saccadic vector lines point_cloud.png: discrete spatial distribution of all recorded gaze coordinate points over the stimulus. [...]
Paper publication
Paper publicationSeptember 24, 2025Our paper: Towards Zero-Shot ISO/ICAO Face Compliance Verification via CLIP-IQA and Natural Language PromptingN. Di Domenico, G. Borghi, A. Franco, D. Maltoni has been published at the International Joint Conference on Biometrics 2025 (8-11 September, Osaka, Japan)! Take a look at it now! [...]
An Overview of the Bologna Online Evaluation Platform- Lunch Talk @ EAB
An Overview of the Bologna Online Evaluation Platform- Lunch Talk @ EABJanuary 23, 2024On Tuesday, January 23rd, Lorenzo Pellegrini will present the Bologna Online Evaluation Platform (BOEP) at the Lunch Talk session of the European Association for Biometrics (EAB). During the talk, titled “Evaluating Morphing Attack Detectors: An Overview of the Bologna Online Evaluation Platform”, the BOEP and FVC-ongoing platform will be briefly described and the main steps needed to submit an algorithm for evaluation will be shown. The goal of the Bologna Online Evaluation Platform is to provide a trusted platform that can be used to evaluate the performance of face morphing attack detection (MAD) algorithms.​ The platform can be used by academic groups, industry actors, and independent developers to evaluate the performance of both Single Image and Differential attack detectors on various benchmarks. Tests are carried out on sequestered datasets in a controlled environment​. The platform was recently updated to allow participants to use a more modern submission format based on Python scripts and Docker environments. Event link: https://eab.org/events/program/334?ts=1706021826469Talk slides: http://miatbiolab.csr.unibo.it/wp-content/uploads/2024/01/BOEP-EAB-Lunch-Talk-2024-01-23.pdf EAB members will be able to access the full registration of the talk. [...]
NVIDIA AI visits Cesena campus
NVIDIA AI visits Cesena campusJanuary 25, 2023On Thursday, January 26th, at 11 AM, Giuseppe Fiameni (NVIDIA) and Andrea Pilzer (NVIDIA) will give an invited talk on “GPU-Accelerated Computing”. Giuseppe Fiameni is a Solution Architect and Data Scientist at NVIDIA, where he oversees the NVIDIA AI Technology Center in Italy, a collaboration among NVIDIA, CINI and CINECA to accelerate academic research in the field of Artificial Intelligence. Andrea Pilzer is a Solution Architect at NVIDIA and works with NVIDIA AI Technology Center in Italy. He was postdoc at Aalto University working on uncertainty estimation for deep learning, worked at Huawei Ireland and got his Ph.D. The seminar will be held at Aula 2.6, Cesena Campus, Via dell’Università 50 – Cesena. More info on the seminar is in the flier. [...]
The Recycling Symbols dataset
The Recycling Symbols datasetNovember 29, 2022We are releasing the dataset of recycling symbols used in the paper “A weakly supervised approach for recycling code recognition”, which will be published on Expert Systems with Applications soon! 🥳 The dataset is a simple collection of images depicting the recycling symbols commonly found in consumer products. This is similar to the Seven Plastics dataset released on Kaggle here, but it includes non-plastic symbols (FE, GL, …) and many more temporally correlated images. The dataset is partially annotated with the bounding boxes of the triangles and the text elements depicted in the images. In our paper, we describe how a weakly supervised pipeline can create missing boxes and then use them in an iterative training scheme. The dataset also carries a set of negative examples obtained by cropping the frames from the original videos in a part of the image far from the triangle. More info on our paper is coming soon :-)! Dataset download: here [...]
Invited talk on “AI Etichs”.
Invited talk on “AI Etichs”.November 29, 2022On Wednesday, November 30th, at 4:00 PM, Gabriele Graffieti will give an invited talk on “AI Etichs“.Gabriele Graffieti is Senior Computer Vision Algorithm Engineer at VisLab (an Ambarella Inc. company, located in Parma (Italy).The seminar will be held at Aula 4.2 CAD, DISI Department, Cesena, and it is part of the Machine Learning and Data Mining course (Module 2, prof. Guido Borghi). [...]
Invited talk on “AI Solutions for Real-World Business”
Invited talk on “AI Solutions for Real-World Business”November 29, 2022On Wednesday, November 23rd, at 2:00 PM, Cosimo Fiorini will give an invited talk on “AI Solutions for Real-World Business“. Cosimo Fiorini is Head of Data Science at Ammagamma company, located in Modena (Italy). The seminar will be held at Aula 4.2 CAD, DISI Department, Cesena, and it is part of the Machine Learning and Data Mining course (Module 2, prof. Guido Borghi). [...]
Cognitive Robots 2022
Cognitive Robots 2022May 30, 2022Our laboratory has presented the research work titled Continual Learning for Real Life Robotics at the event “Cognitive Robots 2022” held in conjunction with the International Conference on Image Analysis and Processing (ICIAP), Lecce, Italy. [...]
Paper publication
Paper publicationMay 18, 2022Our paper: Incremental Training of Face Morphing DetectorsG. Borghi, G. Graffieti, A. Franco, D. Maltoni has been accepted in the 26th International Conference on Pattern Recognition (August 21-25, 2022 – Montréal Québec)! [...]
Paper publication
Paper publicationMay 5, 2022Our paper: Continual Learning in Real-Life Applications G. Graffieti, G. Borghi and D. Maltoni has been accepted for publication in the IEEE Robotics and Automation Letters journal! Check it out on IEEE Explore! [...]
They’re talking about us!
They’re talking about us!December 17, 2021Resto del Carlino (local edition) [link to the article (italian)] Unibo Magazine [link to the article (italian)] [...]
1 place in the ICCV 2021 SSLAD competition!
1 place in the ICCV 2021 SSLAD competition!November 22, 2021The Unibo team coordinated by prof. Davide Maltoni at the UOS DISI in Cesena, which includes: Gabriele Graffieti (PhD student Data Science), Guido Borghi (RTD-A DISI) and Matteo Ferrara (Associate Professor DISI), won the competition Continual learning – supervided classification organized within the workshop Self-supervised learning for next-generation industry-level Autonomous Driving (ICCV 2021) which was attended by 47 teams from all over the world. [...]
Hello, World! Welcome to the new MI@Biolab website!
Hello, World! Welcome to the new MI@Biolab website!October 2, 2021This is the first test news of the MI@Biolab website! Here, you will find updates and news from the laboratory and the Machine Learning community. Stay tuned! [...]