Many practical applications require training of semantic segmentation models on unlabelled datasets and their execution on low-resource hardware. Distillation from a trained source model may represent a solution for the first but does not account for the different distribution of the training data. Unsupervised domain adaptation (UDA) techniques claim to solve the domain shift, but in most cases assume the availability of the source data or an accessible white-box source model, which in practical applications are often unavailable for commercial and/or safety reasons. In this paper, we investigate a more challenging setting in which a lightweight model has to be trained on a target unlabelled dataset for semantic segmentation, under the assumption that we have access only to black-box source model predictions. Our method, named CoRTe, consists of (i) a pseudo-labelling function that extracts reliable knowledge from the black-box source model using its relative confidence, (ii) a pseudo label refinement method to retain and enhance the novel information learned by the student model on the target data, and (iii) a consistent training of the model using the extracted pseudo labels. We benchmark CoRTe on two synthetic-to-real settings, demonstrating remarkable results when using black-box models to transfer knowledge on lightweight models for a target data distribution.

Cross-Domain Transfer Learning with CoRTe: Consistent and Reliable Transfer from Black-Box to Lightweight Segmentation Model / Cuttano, Claudia; Tavera, Antonio; Cermelli, Fabio; Averta, GIUSEPPE BRUNO; Caputo, Barbara. - (2023), pp. 1404-1414. (Intervento presentato al convegno IEEE/CVF International Conference on Computer Vision (ICCV) Workshops tenutosi a Paris (FR) nel 02-06 October 2023) [10.1109/ICCVW60793.2023.00153].

Cross-Domain Transfer Learning with CoRTe: Consistent and Reliable Transfer from Black-Box to Lightweight Segmentation Model

Cuttano Claudia;Tavera Antonio;Cermelli Fabio;Averta Giuseppe;Caputo Barbara
2023

Abstract

Many practical applications require training of semantic segmentation models on unlabelled datasets and their execution on low-resource hardware. Distillation from a trained source model may represent a solution for the first but does not account for the different distribution of the training data. Unsupervised domain adaptation (UDA) techniques claim to solve the domain shift, but in most cases assume the availability of the source data or an accessible white-box source model, which in practical applications are often unavailable for commercial and/or safety reasons. In this paper, we investigate a more challenging setting in which a lightweight model has to be trained on a target unlabelled dataset for semantic segmentation, under the assumption that we have access only to black-box source model predictions. Our method, named CoRTe, consists of (i) a pseudo-labelling function that extracts reliable knowledge from the black-box source model using its relative confidence, (ii) a pseudo label refinement method to retain and enhance the novel information learned by the student model on the target data, and (iii) a consistent training of the model using the extracted pseudo labels. We benchmark CoRTe on two synthetic-to-real settings, demonstrating remarkable results when using black-box models to transfer knowledge on lightweight models for a target data distribution.
2023
979-8-3503-0744-3
File in questo prodotto:
File Dimensione Formato  
Cuttano_Cross-Domain_Transfer_Learning_with_CoRTe_Consistent_and_Reliable_Transfer_from_ICCVW_2023_paper.pdf

accesso aperto

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 1.47 MB
Formato Adobe PDF
1.47 MB Adobe PDF Visualizza/Apri
Cross-Domain_Transfer_Learning_with_CoRTe_Consistent_and_Reliable_Transfer_from_Black-Box_to_Lightweight_Segmentation_Model.pdf

non disponibili

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 1.76 MB
Formato Adobe PDF
1.76 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2982904