MétaCan
Menu
Back to cohort

Data Augmentation Methods for Low Resolution Facial Images

2022· article· en· W4312102546 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenueTENCON 2022 - 2022 IEEE Region 10 Conference (TENCON) · 2022
Typearticle
Languageen
FieldComputer Science
TopicAdvanced Image Processing Techniques
Canadian institutionsUniversity of Windsor
Fundersnot available
KeywordsOverfittingComputer scienceArtificial intelligenceRegularization (linguistics)Pattern recognition (psychology)Face (sociological concept)Set (abstract data type)Data setTraining setResolution (logic)Image resolutionMachine learningData miningArtificial neural network

Abstract

fetched live from OpenAlex

Regularization techniques are useful in alleviating the problem of overfitting in super-resolution using deep learning. In this paper, multiple kinds of augmentation techniques such as CutOut, CutMix, Cutblur, Blend and Mix of augmentation techniques (MOA) are applied to CELEBA database. Both MOA and CutBlur methods improve the performance of face super resolution method. The first set of experiments are conducted without applying augmentation methods. Next set of experiments are conducted applying both MOA augmentation and CutOut, CutMix, CutBlur, Blend methods. The experiments are carried out for different values of patch size and CutBlur ratio. CELEBA dataset is used for the experiment. The effectiveness of the application of data augmentation technique is tested in EDSR model. The results shows that data augmentation techniques improve the performance of super resolution methods. CutBlur technique performs the best among all the methods.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.002
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow), Science and technology studies
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Other design · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: Methods
Teacher disagreement score0.761
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0020.001
Meta-epidemiology (narrow)0.0000.001
Meta-epidemiology (broad)0.0010.000
Bibliometrics0.0000.001
Science and technology studies0.0010.000
Scholarly communication0.0000.003
Open science0.0050.002
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.114
GPT teacher head0.403
Teacher spread0.288 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it