Deep Learning Techniques for Visual Counting

TL;DR

This paper advances deep learning for visual counting by addressing data scarcity, domain adaptation, weak supervision, and deploying efficient models on embedded systems, improving scalability and real-world applicability.

Contribution

It introduces datasets from virtual environments, domain adaptation strategies, weakly supervised counting methods, and efficient models for embedded devices, enhancing deep learning's robustness and deployment.

Findings

Datasets from virtual environments improve training data availability.

Domain adaptation reduces performance gap between training and real data.

Embedded solutions enable real-time counting on resource-constrained devices.

Abstract

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.