FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4317038436> ?p ?o ?g. }

• W4317038436 endingPage "17" @default.
• W4317038436 startingPage "1" @default.
• W4317038436 abstract "Domain adaptive semantic segmentation attempts to make satisfactory dense predictions on an unlabeled target domain by utilizing the supervised model trained on a labeled source domain. One popular solution is self-training, which retrains the model with pseudo labels on target instances. Plenty of approaches tend to alleviate noisy pseudo labels, however, they ignore the intrinsic connection of the training data, i.e., intra-class compactness and inter-class dispersion between pixel representations across and within domains. In consequence, they struggle to handle cross-domain semantic variations and fail to build a well-structured embedding space, leading to less discrimination and poor generalization. In this work, we propose emantic-Guided Pixel Contrast (SePiCo), a novel one-stage adaptation framework that highlights the semantic concepts of individual pixels to promote learning of class-discriminative and class-balanced pixel representations across domains, eventually boosting the performance of self-training methods. Specifically, to explore proper semantic concepts, we first investigate a centroid-aware pixel contrast that employs the category centroids of the entire source domain or a single source image to guide the learning of discriminative features. Considering the possible lack of category diversity in semantic concepts, we then blaze a trail of distributional perspective to involve a sufficient quantity of instances, namely distribution-aware pixel contrast, in which we approximate the true distribution of each semantic category from the statistics of labeled source data. Moreover, such an optimization objective can derive a closed-form upper bound by implicitly involving an infinite number of (dis)similar pairs, making it computationally efficient. Extensive experiments show that SePiCo not only helps stabilize training but also yields discriminative representations, making significant progress on both synthetic-to-real and daytime-to-nighttime adaptation scenarios. The code and models are available at https://github.com/BIT-DA/SePiCo." @default.
• W4317038436 created "2023-01-18" @default.
• W4317038436 creator A5009558549 @default.
• W4317038436 creator A5013240918 @default.
• W4317038436 creator A5035319531 @default.
• W4317038436 creator A5054991337 @default.
• W4317038436 creator A5072070557 @default.
• W4317038436 creator A5073287670 @default.
• W4317038436 date "2023-01-01" @default.
• W4317038436 modified "2023-10-13" @default.
• W4317038436 title "SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation" @default.
• W4317038436 cites W1903029394 @default.
• W4317038436 cites W2108598243 @default.
• W4317038436 cites W2138621090 @default.
• W4317038436 cites W2150066425 @default.
• W4317038436 cites W2165698076 @default.
• W4317038436 cites W2194775991 @default.
• W4317038436 cites W2340897893 @default.
• W4317038436 cites W2412782625 @default.
• W4317038436 cites W2431874326 @default.
• W4317038436 cites W2487365028 @default.
• W4317038436 cites W2560023338 @default.
• W4317038436 cites W2563705555 @default.
• W4317038436 cites W2593768305 @default.
• W4317038436 cites W2744813330 @default.
• W4317038436 cites W2887964057 @default.
• W4317038436 cites W2889945482 @default.
• W4317038436 cites W2895281799 @default.
• W4317038436 cites W2947263797 @default.
• W4317038436 cites W2952735550 @default.
• W4317038436 cites W2963073217 @default.
• W4317038436 cites W2963107255 @default.
• W4317038436 cites W2964115968 @default.
• W4317038436 cites W2968634921 @default.
• W4317038436 cites W2972285644 @default.
• W4317038436 cites W2985406498 @default.
• W4317038436 cites W2985409929 @default.
• W4317038436 cites W3000172657 @default.
• W4317038436 cites W3002450089 @default.
• W4317038436 cites W3010381534 @default.
• W4317038436 cites W3034417116 @default.
• W4317038436 cites W3034562924 @default.
• W4317038436 cites W3035236545 @default.
• W4317038436 cites W3035294798 @default.
• W4317038436 cites W3035524453 @default.
• W4317038436 cites W3035564946 @default.
• W4317038436 cites W3040318838 @default.
• W4317038436 cites W3108560336 @default.
• W4317038436 cites W3108566666 @default.
• W4317038436 cites W3110011650 @default.
• W4317038436 cites W3114677757 @default.
• W4317038436 cites W3119635706 @default.
• W4317038436 cites W3120562181 @default.
• W4317038436 cites W3120804725 @default.
• W4317038436 cites W3122412340 @default.
• W4317038436 cites W3138516171 @default.
• W4317038436 cites W3157653192 @default.
• W4317038436 cites W3164066238 @default.
• W4317038436 cites W3168822201 @default.
• W4317038436 cites W3170700905 @default.
• W4317038436 cites W3172615411 @default.
• W4317038436 cites W3175294391 @default.
• W4317038436 cites W3175308890 @default.
• W4317038436 cites W3176820334 @default.
• W4317038436 cites W3195315815 @default.
• W4317038436 cites W3202422321 @default.
• W4317038436 cites W3204469733 @default.
• W4317038436 cites W3206154535 @default.
• W4317038436 cites W3210218433 @default.
• W4317038436 cites W3213646008 @default.
• W4317038436 cites W3216480618 @default.
• W4317038436 cites W3217147624 @default.
• W4317038436 cites W4229723800 @default.
• W4317038436 cites W4287124998 @default.
• W4317038436 cites W639708223 @default.
• W4317038436 doi "https://doi.org/10.1109/tpami.2023.3237740" @default.
• W4317038436 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37819799" @default.
• W4317038436 hasPublicationYear "2023" @default.
• W4317038436 type Work @default.
• W4317038436 citedByCount "15" @default.
• W4317038436 countsByYear W43170384362023 @default.
• W4317038436 crossrefType "journal-article" @default.
• W4317038436 hasAuthorship W4317038436A5009558549 @default.
• W4317038436 hasAuthorship W4317038436A5013240918 @default.
• W4317038436 hasAuthorship W4317038436A5035319531 @default.
• W4317038436 hasAuthorship W4317038436A5054991337 @default.
• W4317038436 hasAuthorship W4317038436A5072070557 @default.
• W4317038436 hasAuthorship W4317038436A5073287670 @default.
• W4317038436 hasBestOaLocation W43170384362 @default.
• W4317038436 hasConcept C134306372 @default.
• W4317038436 hasConcept C146599234 @default.
• W4317038436 hasConcept C153180895 @default.
• W4317038436 hasConcept C154945302 @default.
• W4317038436 hasConcept C160633673 @default.
• W4317038436 hasConcept C2776502983 @default.
• W4317038436 hasConcept C33923547 @default.
• W4317038436 hasConcept C36503486 @default.
• W4317038436 hasConcept C41008148 @default.

• next