Pith. sign in

REVIEW 12 cited by

Revisiting Image Aesthetic Assessment via Self-Supervised Feature Learning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1911.11419 v1 pith:GQIXBVUE submitted 2019-11-26 cs.CV

classification cs.CV
keywords aestheticimageassessmentfeaturepretextbenchmarkdatasetsfeatures
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Visual aesthetic assessment has been an active research field for decades. Although latest methods have achieved promising performance on benchmark datasets, they typically rely on a large number of manual annotations including both aesthetic labels and related image attributes. In this paper, we revisit the problem of image aesthetic assessment from the self-supervised feature learning perspective. Our motivation is that a suitable feature representation for image aesthetic assessment should be able to distinguish different expert-designed image manipulations, which have close relationships with negative aesthetic effects. To this end, we design two novel pretext tasks to identify the types and parameters of editing operations applied to synthetic instances. The features from our pretext tasks are then adapted for a one-layer linear classifier to evaluate the performance in terms of binary aesthetic classification. We conduct extensive quantitative experiments on three benchmark datasets and demonstrate that our approach can faithfully extract aesthetics-aware features and outperform alternative pretext schemes. Moreover, we achieve comparable results to state-of-the-art supervised methods that use 10 million labels from ImageNet.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 142 citations worldwide. Full citation record

  1. Fast Isotropic Median Filtering

    cs.CV 2025-05 conditional novelty 8.0 of 10

    A new 2D 'omnigram' data structure enables exact median filtering with arbitrary convex kernels, including circles, at state-of-the-art speeds.

  2. UAV3DCrop: Benchmarking 3D Reconstruction in Repeated Multi-Angle UAV Crop Surveys

    cs.CV 2026-08 conditional novelty 6.0 of 10

    Current 3D reconstruction methods are not interchangeable for crop monitoring: appearance, geometry, and canopy-height rankings diverge, and most zero-shot models fail at metric scale.

  3. Quadrature-Aware Complex-Linear Neural Operator for Boundary-to-Field Prediction in Resonant Acoustics

    physics.flu-dyn 2026-07 conditional novelty 6.0 of 10

    A quadrature-aware complex-linear neural operator halves field error versus DeepONet and enforces exact source superposition for resonant cavity acoustics.

  4. SARe: Structure-Aware Generative 3D Fragment Reassembly

    cs.CV 2026-03 conditional novelty 6.0 of 10

    SARe improves many-fragment 3D reassembly by jointly predicting fracture-surface labels and a contact graph during flow-based pose generation, plus inference-time resampling of uncertain regions.

  5. A Fast Parallel Median Filtering Algorithm Using Hierarchical Tiling

    cs.DC 2025-07 conditional novelty 6.0 of 10

    Hierarchical tiling lets sorting-based median filters on GPUs reach O(k log k) and O(k) per-pixel complexity for k x k kernels, with large measured speedups.

  6. Symmetries of weighted networks: weight approximation method and its application to food webs

    physics.soc-ph 2025-06 conditional novelty 6.0 of 10

    Logarithmic binning of edge weights turns weighted food webs into graphs where automorphisms appear, and the resulting orbits are almost always of size two or three.

  7. Scalable and High-Quality Neural Implicit Representation for 3D Reconstruction

    cs.CV 2025-01 conditional novelty 6.0 of 10

    A scene is reconstructed as a graph of overlapping local neural SDFs that are registered and blended, improving detail and enabling large-scale reconstruction.

  8. Iterating the Transient Light Transport Matrix for Non-Line-of-Sight Imaging

    physics.optics 2024-12 conditional novelty 6.0 of 10

    A SPAD array captures the full first-order transient light transport matrix of a relay wall, and beamforming algorithms extract the second-order transient light transport matrix of the hidden scene, enabling relightin...

  9. Real-Time Global Illumination Decomposition of Videos

    cs.CV 2019-08 conditional novelty 6.0 of 10

    A monocular RGB video can be decomposed in real time into reflectance, direct illumination, and multiple indirect illumination layers using sparse base colors and a customized GPU optimizer.

  10. Fitting Spherical Gaussians to Dynamic HDRI Sequences

    cs.CV 2024-12 conditional novelty 5.0 of 10

    A method that fits dynamic HDRI lighting sequences with anisotropic spherical Gaussians using L1, diffuse, and temporal consistency losses.

  11. Tangi: a Tool to Create Tangible Artifacts for Sharing Insights from 360$^\circ$ Video

    cs.HC 2024-11 conditional novelty 5.0 of 10

    The paper introduces Tangi, a tool that converts 360-degree video frames into flat and polyhedral paper artifacts, and reports an initial qualitative study suggesting these artifacts support collaborative design analysis.

  12. Exploring Flexible Scenario Generation in Godot Simulator

    cs.AI 2024-12 reject novelty 4.0 of 10

    A prototype pipeline reconstructs road scenes in the Godot game engine from images, with an unvalidated STL-based method to constrain road modifications.

Pith tools