Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 20 of 33 for “"Dense Prediction"”.

  1. Continual Learning for Deep Dense Prediction

    … alleviate catastrophic forgetting for pixel-wise dense labeling problems. We build upon the regularization technique using knowledge distillation to minimize the discrepancy between the posterior distribution of pixel class labels for old tasks predicted from 1) the original and 2) the updated …

    vt Repository record for Continual Learning for Deep Dense Prediction (opens in a new tab)

  2. Learning dense prediction: from correspondence to segmentation

    <p>Dense prediction is the task of predicting a label for each pixel in the image. Given 3D data (point clouds or RGB-D images) as input, dense prediction can also be extended to 3D space and assign each 3D point/location a label. According to the label type, dense prediction can be mainly …

    oxford Repository record for Learning dense prediction: from correspondence to segmentation (opens in a new tab)

  3. Efficient Fully Convolutional Networks for Dense Prediction Tasks

    Dense prediction is a family of fundamental problems in computer vision, which learns a mapping from input images to complex output structures, including semantic segmentation, depth estimation, and object detection, among many others. Pixel-level labeling is required in such tasks. Deep neural …

    adelaide Repository record for Efficient Fully Convolutional Networks for Dense Prediction Tasks (opens in a new tab)

  4. Learnable Structured Attention For Student-Aware Knowledge Distillation In Dense Prediction

    Dense prediction tasks, including object detection and semantic segmentation, require models to produce structured predictions and are widely used in real-world vision applications. Although deep networks have achieved strong performance on these tasks, their high computational cost limits …

    uab Repository record for Learnable Structured Attention For Student-Aware Knowledge Distillation In Dense Prediction (opens in a new tab)

  5. Studies of lighting for dense prediction and generation in computer vision

    Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2024-09-16 without embargo terms

    uiuc Repository record for Studies of lighting for dense prediction and generation in computer vision (opens in a new tab)

  6. Usable Machine Learning for Remote Sensing Data

    … (ML) and remote sensing data is a set of dense predictions that form a predicted map for a geographic region. However, most prior work involving ML and remote sensing follows the traditional practice of reporting metrics on a set of independent, geographically-sparse samples and does not …

    maryland Repository record for Usable Machine Learning for Remote Sensing Data (opens in a new tab)

  7. Deep Object Segmentation and Beyond

    … and image matting. Third, we propose dense contrastive learning (DenseCL) to learn better representation from large-scale unlabeled images for dense prediction tasks such as segmentation. The proposed DenseCL performs dense pairwise contrastive learning at the level of pixels. Our …

    adelaide Repository record for Deep Object Segmentation and Beyond (opens in a new tab)

  8. Mining Visual Knowledge from Pre-trained Models

    … downstream tasks such as recognition, dense prediction, 3D generation, and reference-based image completion.

    cornell Repository record for Mining Visual Knowledge from Pre-trained Models (opens in a new tab)

  9. Towards Effective and Efficient Semantic Segmentation

    Semantic segmentation, a crucial per-pixel dense prediction task, requires both accuracy and efficiency for effective application in real-world scenarios. This thesis introduces a series of novel methods that address the challenges of semantic segmentation, focusing on achieving high performance …

    adelaide Repository record for Towards Effective and Efficient Semantic Segmentation (opens in a new tab)

  10. An evaluation of deep learning semantic segmentation for land cover classification of oblique ground-based photography

    … the application of deep learning methods for the dense prediction of land cover types in oblique ground-based photography. While deep learning approaches are widely used in land cover classification of remote-sensing data (i.e., aerial and satellite orthoimagery) for change detection analysis, …

    uvic Repository record for An evaluation of deep learning semantic segmentation for land cover classification of oblique ground-based photography (opens in a new tab)

  11. Learning efficient representations for generalizable segmentation in data-limited scenarios

    … minimal supervision is important, especially for dense prediction tasks like segmentation, where collecting pixel-level annotations is costly or impractical. These challenges are greater in data-limited domains and under distribution shifts, requiring models that can generalize with limited …

    ubc Repository record for Learning efficient representations for generalizable segmentation in data-limited scenarios (opens in a new tab) Download PDF for Learning efficient representations for generalizable segmentation in data-limited scenarios (opens in a new tab)

  12. Efficient processing of temporally asynchronous and spatially unstructured event-based camera data

    … framework tailored specifically for event-based dense prediction tasks, which require precise pixel-level associations. Our method uniquely manages sparse and asynchronous event streams by projecting them into hierarchical, multi-level latent memory structures through an attention-based spatial …

    gatech Repository record for Efficient processing of temporally asynchronous and spatially unstructured event-based camera data (opens in a new tab)

  13. Methods to safely scale modern neural networks

    … don’t know. Neural networks can make confident predictions even when they are incorrect with potentially catastrophic consequences in safety-critical downstream tasks like medical diagnosis and autonomous driving. Second, in the area of scalability, deep learning remains a data and …

    oxford Repository record for Methods to safely scale modern neural networks (opens in a new tab)

  14. Deep Learning for Multipitch Detection and Melody Extraction

    … in music information retrieval. They are both dense prediction tasks similar to semantic image segmentation. However, unlike image segmentation where the constituent parts of an object often exist in a compact region, in both tasks the pitches and harmonics of a music note are …

    adelaide Repository record for Deep Learning for Multipitch Detection and Melody Extraction (opens in a new tab)

  15. Efficient Fully-Convolutional Networks for Image Perception

    … their sim ple structures. They perform well on dense prediction tasks such as semantic segmentation, where the networks consist of a backbone module for visual feature extraction and a task-specific module for result generation. Designing the task-specific modules helps us understand how these …

    adelaide Repository record for Efficient Fully-Convolutional Networks for Image Perception (opens in a new tab)

  16. Physical-World Visual Intelligence Through Structured and Generative Representations

    … producing scalable perception backbones for dense prediction, vision-language alignment, and multimodal understanding. The second part focuses on structured reasoning about the physical world. I introduce a framework that extracts explicit 3D representations from visual inputs and encodes …

    jhu Repository record for Physical-World Visual Intelligence Through Structured and Generative Representations (opens in a new tab)

Page 1 of 2