Human-centered deep compositional model for handling occlusions

Koporec, Gregor (author), Perš, Janez (author)

Despite their powerful discriminative abilities, Convolutional Neural Networks (CNNs) lack the properties of generative models. This leads to a decreased performance in environments where objects are poorly visible. Solving such a problem by adding more training samples can quickly lead to a combinatorial explosion, therefore the underlying architecture has to be changed instead. This work proposes a Human-Centered Deep Compositional model (HCDC) that combines low-level visual discrimination of a CNN and the high-level reasoning of a Hierarchical Compositional model (HCM). Defined as a transparent model, it can be optimized to real-world environments by adding compactly encoded domain knowledge from human studies and physical laws. The new FridgeNetv2 dataset and a mixture of publicly available datasets are used as a benchmark. The experimental results show the proposed model is explainable, has higher discriminative and generative power, and better handles the occlusion than the current state-of-the-art Mask-RCNN in instance segmentation tasks.

• Language: English
• Keywords: computer vision, deep learning, convolutional neural networks, hierarchical compositional model, occlusion, discriminability, generalizability, interpretability, domain knowledge, instance segmentation, occlusion handling
• Work type: Article
• Typology: 1.01 - Original Scientific Article
• Organization: FE - Faculty of Electrical Engineering
• Publication status: Published
• Publication version: Version of Record
• Year: 2023
• Number of pages: 14 str.
• Numbering: Vol. 138, art. 109397
• PID: 20.500.12556/RUL-148675
• UDC: 004
• ISSN on article: 0031-3203
• DOI: 10.1016/j.patcog.2023.109397
• COBISS.SI-ID: 142438403
• Publication date in RUL: 29.08.2023

Record is a part of a journal

• Title: Pattern recognition
• Shortened title: Pattern recogn.
• Publisher: Elsevier
• ISSN: 0031-3203
• COBISS.SI-ID: 26103040

Licences

• License: CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
• Link: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Description: The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.

Secondary language

• Language: Slovenian
• Keywords: računalniški vid, globoko učenje, konvolucijske nevronske mreže, hierarhični kompozicionalni model, zakrivanje, interpretabilnost, poznavanje področja