Publications

2024

1. Capsule Networks Do Not Need to Model Everything

   Riccardo Renzulli, Enzo Tartaglione, and Marco Grangetto

   2024

   In review
2. Unsupervised Contrastive Analysis for Salient Pattern Detection using Conditional Diffusion Models

   Cristiano Patrício, Carlo Alberto Barbano, Riccardo Renzulli, and 4 more authors

   2024

   In review
3. Boost Your NeRF: A Model-Agnostic Mixture of Experts Framework for High Quality and Efficient Rendering

   Francesco Di Sario, Riccardo Renzulli, Enzo Tartaglione, and 1 more author

   2024

   Accepted at the 18th European Conference on Computer Vision (ECCV)
4. AI-Assisted Diagnosis for Covid-19 CXR Screening: from Data Collection to Clinical Validation

   Carlo Alberto Barbano, Riccardo Renzulli, Marco Grosso, and 3 more authors

   In 2024 IEEE International Symposium on Biomedical Imaging (ISBI), 2024

   DOI
5. Hierarchical Object-Centric Learning with Capsule Networks

   Riccardo Renzulli

   2024

   PhD thesis

2023

1. LSVL: Large-scale season-invariant visual localization for UAVs

   Jouko Kinnari, Riccardo Renzulli, Francesco Verdoja, and 1 more author

   Robotics and Autonomous Systems, 2023

   Abs DOI

   Localization of autonomous unmanned aerial vehicles (UAVs) relies heavily on Global Navigation Satellite Systems (GNSS), which are susceptible to interference. Especially in security applications, robust localization algorithms independent of GNSS are needed to provide dependable operations of autonomous UAVs also in interfered conditions. Typical non-GNSS visual localization approaches rely on known starting pose, work only on a small-sized map, or require known flight paths before a mission starts. We consider the problem of localization with no information on initial pose or planned flight path. We propose a solution for global visual localization on large maps, based on matching orthoprojected UAV images to satellite imagery using learned season-invariant descriptors, and test with environment sizes up to 100 km2. We show that the method is able to determine heading, latitude and longitude of the UAV at 12.6–18.7 m lateral translation error in as few as 23.2–44.4 updates from an uninformed initialization, also in situations of significant seasonal appearance difference (winter–summer) between the UAV image and the map. We evaluate the characteristics of multiple neural network architectures for generating the descriptors, and likelihood estimation methods that are able to provide fast convergence and low localization error. We also evaluate the operation of the algorithm using real UAV data and evaluate running time on a real-time embedded platform. We believe this is the first work that is able to recover the pose of an UAV at this scale and rate of convergence, while allowing significant seasonal difference between camera observations and map.

2. Shannon Strikes Again! Entropy-Based Pruning in Deep Neural Networks for Transfer Learning Under Extreme Memory and Computation Budgets

   Gabriele Spadaro, Riccardo Renzulli, Andrea Bragagnolo, and 4 more authors

   In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, Oct 2023

   Abs

   Deep neural networks have become the de-facto standard across various computer science domains. Nonetheless, effectively training these deep networks remains challenging and resource-intensive. This paper investigates the efficacy of pruned deep learning models in transfer learning scenarios under extremely low memory budgets, tailored for TinyML models. Our study reveals that the source task’s model with the highest activation entropy outperforms others in the target task. Motivated by this, we propose an entropy-based Efficient Neural Transfer with Reduced Overhead via PrunIng (ENTROPI) algorithm. Through comprehensive experiments on diverse models (ResNet18 and MobileNet-v3) and target datasets (CIFAR-100, VLCS, and PACS), we substantiate the superior generalization achieved by transfer learning from the entropy-pruned model. Quantitative measures for entropy provide valuable insights into the reasons behind the observed performance improvements. The results underscore ENTROPI’s potential as an efficient solution for enhancing generalization in data-limited transfer learning tasks.

3. Two is Better than One: Achieving High-Quality 3D Scene Modeling with a NeRF Ensemble

   Francesco Di Sario, Riccardo Renzulli, Enzo Tartaglione, and 1 more author

   In Image Analysis and Processing – ICIAP 2023, Oct 2023

   Abs Code

   Neural Radiance Field (NeRF) is a popular method for synthesizing novel views of a scene from a set of input images. While NeRF has demonstrated state-of-the-art performance in several applications, it suffers from high computational requirements. Recent works have attempted to address these issues by including explicit volumetric information, which makes the optimization process difficult when fine-graining the voxel grids. In this paper, we propose an ensemble approach that combines the strengths of two NeRF models to achieve superior results compared to state-of-the-art architectures, with a similar number of parameters. Experimental results show that our ensemble approach is a promising strategy for performance enhancement, and beats vanilla approaches under the same parameter’s cardinality constraint.

4. Neural network-derived perfusion maps: A model-free approach to computed tomography perfusion in patients with acute ischemic stroke

   Umberto A. Gava, Federico D’Agata, Enzo Tartaglione, and 8 more authors

   Frontiers in Neuroinformatics, Oct 2023

   Abs DOI Code

   ObjectiveIn this study, we investigate whether a Convolutional Neural Network (CNN) can generate informative parametric maps from the pre-processed CT perfusion data in patients with acute ischemic stroke in a clinical setting.MethodsThe CNN training was performed on a subset of 100 pre-processed perfusion CT dataset, while 15 samples were kept for testing. All the data used for the training/testing of the network and for generating ground truth (GT) maps, using a state-of-the-art deconvolution algorithm, were previously pre-processed using a pipeline for motion correction and filtering. Threefold cross validation had been used to estimate the performance of the model on unseen data, reporting Mean Squared Error (MSE). Maps accuracy had been checked through manual segmentation of infarct core and total hypo-perfused regions on both CNN-derived and GT maps. Concordance among segmented lesions was assessed using the Dice Similarity Coefficient (DSC). Correlation and agreement among different perfusion analysis methods were evaluated using mean absolute volume differences, Pearson correlation coefficients, Bland-Altman analysis, and coefficient of repeatability across lesion volumes.ResultsThe MSE was very low for two out of three maps, and low in the remaining map, showing good generalizability. Mean Dice scores from two different raters and the GT maps ranged from 0.80 to 0.87. Inter-rater concordance was high, and a strong correlation was found between lesion volumes of CNN maps and GT maps (0.99, 0.98, respectively).ConclusionThe agreement between our CNN-based perfusion maps and the state-of-the-art deconvolution-algorithm perfusion analysis maps, highlights the potential of machine learning methods applied to perfusion analysis. CNN approaches can reduce the volume of data required by deconvolution algorithms to estimate the ischemic core, and thus might allow the development of novel perfusion protocols with lower radiation dose deployed to the patient.

2022

1. Towards Efficient Capsule Networks

   Riccardo Renzulli, and Marco Grangetto

   In 2022 IEEE International Conference on Image Processing (ICIP), Oct 2022

   Abs DOI Code

   From the moment Neural Networks dominated the scene for image processing, the computational complexity needed to solve the targeted tasks skyrocketed: against such an unsustainable trend, many strategies have been developed, ambitiously targeting performance’s preservation. Promoting sparse topologies, for example, allows the deployment of deep neural networks models on embedded, resource-constrained devices. Recently, Capsule Networks were introduced to enhance explainability of a model, where each capsule is an explicit representation of an object or its parts. These models show promising results on toy datasets, but their low scalability prevents deployment on more complex tasks. In this work, we explore sparsity besides capsule representations to improve their computational efficiency by reducing the number of capsules. We show how pruning with Capsule Network achieves high generalization with less memory requirements, computational effort, and inference and training time.

2. UniToChest: A Lung Image Dataset for Segmentation of Cancerous Nodules on CT Scans

   Hafiza Ayesha Hoor Chaudhry, Riccardo Renzulli, Daniele Perlo, and 8 more authors

   In Image Analysis and Processing – ICIAP 2022, Oct 2022

   Abs Code

   Lung cancer has emerged as a major causes of death and early detection of lung nodules is the key towards early cancer diagnosis and treatment effectiveness assessment. Deep neural networks achieve outstanding results in tasks such as lung nodules detection, segmentation and classification, however their performance depends on the quality of the training images and on the training procedure. This paper introduces UniToChest, a dataset consisting Computed Tomography (CT) scans of 623 patients. Then, we propose a lung nodules segmentation scheme relying on a convolutional neural architecture that we also re-purpose for a nodule detection task. The experimental results show accurate segmentation of lung nodules across a wide diameter range and better detection accuracy over a traditional detection approach. The datasets and the code used in this paper are publicly made available as a baseline reference.

3. Lung Nodules Segmentation with DeepHealth Toolkit

   Hafiza Ayesha Hoor Chaudhry, Riccardo Renzulli, Daniele Perlo, and 7 more authors

   In Image Analysis and Processing. ICIAP 2022 Workshops, Oct 2022

   Abs Code

   The accurate and consistent border segmentation plays an important role in the tumor volume estimation and its treatment in the field of Medical Image Segmentation. Globally, Lung cancer is one of the leading causes of death and the early detection of lung nodules is essential for the early cancer diagnosis and survival rate of patients. The goal of this study was to demonstrate the feasibility of Deephealth toolkit including PyECVL and PyEDDL libraries to precisely segment lung nodules. Experiments for lung nodules segmentation has been carried out on UniToChest using PyECVL and PyEDDL, for data pre-processing as well as neural network training. The results depict accurate segmentation of lung nodules across a wide diameter range and better accuracy over a traditional detection approach. The datasets and the code used in this paper are publicly available as a baseline reference.

2021

1. Capsule Networks with Routing Annealing

   Riccardo Renzulli, Enzo Tartaglione, Attilio Fiandrotti, and 1 more author

   In Artificial Neural Networks and Machine Learning – ICANN 2021, Oct 2021

   Abs

   Capsule Networks overcome some shortcomings of convolutional neural networks organizing neurons into groups of capsules. Capsule layers are dynamically connected by means of an iterative routing mechanism, which models the connection strengths between capsules from different layers. However, whether routing improves the network performance is still object of debate. This work tackles this issue via Routing Annealing (RA), where the number of routing iterations is annealed at training time. This proposal gives some insights on the effectiveness of the routing for Capsule Networks. Our experiments on different datasets and architectures show that RA yields better performance over a reference setup where the number of routing iterations is fixed (even in the limit case with no routing), especially for architectures with fewer parameters.

2017

1. RAT-OWL: Reasoning with Rational Closure in Description Logics of Typicality

   Laura Giordano, Valentina Gliozzi, Gian Luca Pozzato, and 1 more author

   In Joint Proceedings of the 18th Italian Conference on Theoretical Computer Science and the 32nd Italian Conference on Computational Logic, Oct 2017
2. An Efficient Reasoner for Description Logics of Typicality and Rational Closure

   Laura Giordano, Valentina Gliozzi, Gian Luca Pozzato, and 1 more author

   In Proceedings of the 30th International Workshop on Description Logics, Oct 2017