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 12 of 12 for “"Visual prediction"”.

  1. Learning visual tasks with selective attention

    … resulting in significant gains in several visual prediction tasks. We will demonstrate both directly and indirectly supervised models for selecting image regions and show how they can improve performance over baselines by means of focusing on the right areas.

    uiuc Repository record for Learning visual tasks with selective attention (opens in a new tab)

  2. Learning Perceptual Prediction: Learning from Humans and Reasoning about Objects

    … how robots can learn to predict the future from visual observations. We study two main questions. First, how can agents acquire sufficient data to train large prediction models? Second, what structure should we embed into these models? In this work, we address both questions. In order to acquire …

    penn Repository record for Learning Perceptual Prediction: Learning from Humans and Reasoning about Objects (opens in a new tab)

  3. Autoregressive Modeling for Visual Understanding and Generation

    … distribution through a sequence of conditional predictions. Its success in language benefits from discrete semantic tokens and a natural one-dimensional order. Images and videos do not provide an equally canonical sequence: a visual prediction may represent a pixel, region, spatiotemporal …

    jhu Repository record for Autoregressive Modeling for Visual Understanding and Generation (opens in a new tab)

  4. Study of Human Hand-Eye Coordination Using Machine Learning Techniques in a Virtual Reality Setup

    Theories of visually guided action are characterized as closed-loop control in the presence of reliable sources of visual information, and predictive control to compensate for visuomotor delay and temporary occlusion. However, prediction is not well understood. To investigate, a series of studies …

    rit Repository record for Study of Human Hand-Eye Coordination Using Machine Learning Techniques in a Virtual Reality Setup (opens in a new tab)

  5. High Dynamic Range Visual Content Compression

    … research questions of High Dynamic Range (HDR) visual contents compression. The HDR representations are intended to represent the actual physical value of the light rather than exposed value. The current HDR compression schemes are the extension of legacy Low Dynamic Range (LDR) compressions, by …

    whiterose Download PDF for High Dynamic Range Visual Content Compression (opens in a new tab)

  6. Representation of natural image sequences in the primate visual system

    … predicting the future state of the environment. Visual prediction is difficult because the stream of images on the retina evolves according to irregular, curved trajectories. We posit that the visual system seeks to transform natural inputs such that neural representations follow straighter and …

    texas Repository record for Representation of natural image sequences in the primate visual system (opens in a new tab)

  7. Air-Surface Exchange of Persistent Organic Pollutants in North America

    … space maps. These maps allow instant visual prediction of the phase distribution and transport process of a chemical among the three major phases in soil; i.e., air, water and solid. Net volatilization of alpha-hexachlorocyclohexane from water to air was found in the southern Beaufort …

    toronto-retro Repository record for Air-Surface Exchange of Persistent Organic Pollutants in North America (opens in a new tab)

  8. Structured Grounding for Relational Scene Understanding and Reference-Conditioned Generation

    Modern visual models often produce plausible outputs that are not faithful to their evidence or conditions. A scene graph may assign a frequent but incorrect predicate to an entity pair; an image generator may transfer reference appearance to an incorrect region; and a video generator may lose …

    kings Repository record for Structured Grounding for Relational Scene Understanding and Reference-Conditioned Generation (opens in a new tab) Download PDF for Structured Grounding for Relational Scene Understanding and Reference-Conditioned Generation (opens in a new tab)

  9. Human action prediction and forecasting based on visual data

    … focusing on the tasks of vision-based action prediction, action anticipation and next-active-object prediction. Action prediction is defined as the inference of an action label while the action is still ongoing. Such a capability is useful for early response and further action planning. We …

    greece Repository record for Human action prediction and forecasting based on visual data (opens in a new tab)

  10. The impact of visual floor patterns on human locomotion

    … the most fundamental human behaviours, with the visual system guiding our movements through the environment that we are in. Our visual system is thought to have evolved to detect signals necessary for survival, whilst reducing noise, in the environmental niche of our past. Yet, the visual …

    bristol Repository record for The impact of visual floor patterns on human locomotion (opens in a new tab) Download PDF for The impact of visual floor patterns on human locomotion (opens in a new tab)

  11. An investigation of human capability to predict the future location of objects in motion

    Hitting a Major League fastball pitch may be the most difficult task in the sports realm. Anecdotal evidence suggests that certain individuals are able to perform this task reasonably well, perhaps because of superior sensitivity to changes in motion. However, the substantial lack of research …

    gatech Repository record for An investigation of human capability to predict the future location of objects in motion (opens in a new tab)

  12. Of Priors and Particles: Structured and Distributed Approaches to Robot Perception and Control

    … and combining perceptual models for both visual and tactile sensor modalities, common in robotics. Modern variational inference methods will then be examined in the context of online optimization and stochastic optimal control. Specifically, this thesis will contribute (1) data-driven …

    gatech Repository record for Of Priors and Particles: Structured and Distributed Approaches to Robot Perception and Control (opens in a new tab)