ETD

Digital archive of theses discussed at the University of Pisa

 

Thesis etd-04262022-163702


Thesis type
Tesi di dottorato di ricerca
Author
CIAMPI, LUCA
URN
etd-04262022-163702
Thesis title
Deep Learning Techniques for Visual Counting
Academic discipline
ING-INF/05
Course of study
INGEGNERIA DELL'INFORMAZIONE
Supervisors
tutor Dott. Amato, Giuseppe
tutor Prof. Avvenuti, Marco
tutor Dott. Gennaro, Claudio
Keywords
  • domain adaptation
  • counting objects in images
  • computer vision
  • deep learning
  • convolutional neural nertworks
  • synthetic data
Graduation session start date
03/05/2022
Availability
Full
Summary
In this dissertation, we investigated and enhanced Deep Learning (DL) techniques for counting objects, like pedestrians, cells or vehicles, in still images or video frames. In particular, we tackled the challenge related to the lack of data needed for training current DL-based solutions. Given that the budget for labeling is limited, data scarcity still represents an open problem that prevents the scalability of existing solutions based on the supervised learning of neural networks and that is responsible for a significant drop in performance at inference time when new scenarios are presented to these algorithms. We introduced solutions addressing this issue from several complementary sides, collecting datasets gathered from virtual environments automatically labeled, proposing Domain Adaptation strategies aiming at mitigating the domain gap existing between the training and test data distributions, and presenting a counting strategy in a weakly labeled data scenario, i.e., in the presence of non-negligible disagreement between multiple annotators. Moreover, we tackled the non-trivial engineering challenges coming out of the adoption of Convolutional Neural Network-based techniques in environments with limited power resources, introducing solutions for counting vehicles and pedestrians directly onboard embedded vision systems, i.e., devices equipped with constrained computational capabilities that can capture images and elaborate them.
File