A Study on an Automatic System for Analyzing the Facial Beauty of Young Women
Bibliographic record
Abstract
A Study on an Automatic System for Analyzing the Facial Beauty of Young Women \n \nNeha Sultan \n \nBeauty is one of the foremost ideas that define human personality. However, only recently has the concept of beauty been scientifically analyzed. This has mostly been due to extensive research done in the area of face recognition and image processing on identification and classification of human features as contributing to facial beauty. Current research aims at precisely and conclusively understanding how humans classify a given individual's face as beautiful. Due to the lack of published theoretical standards and ground truths for human facial beauty, this is often an ambiguous process. Current methods of analysis and classification of human facial beauty rely mainly on the geometric aspects of human facial beauty. The classifiers used in current research include the k-nearest neighbor algorithm, ridge regression, and basic principal component analysis. \n \nIn this research, various approaches related to the comprehension and analysis of human beauty are presented and the use of these theories is outlined. Each set of theories is translated into a feature model that is tested for classification. Selecting the best set of features that result in the most accurate model for the representation of the human face is a key challenge. This research introduces the combined use of three main groups of features for classification of female facial beauty, to be used with classification through support vector machines. The classifier utilized is Support Vector Machine (SVM) and the accuracy obtained through this classifier is 86%. Current research in the field has produced algorithms with percentages of accuracy that are in the range of 75-85%. The approach used is one of analysis of the central tenets of beauty, the successive application of image processing techniques, and finally the usage of relevant machine learning methods to build an effective system for the automatic assessment of facial beauty. The ground truths used for verifying results are derived from ratings extracted from surveys conducted. \n \n \nThe proposed methodology involves a novel algorithm for the representation of facial beauty, which combines the use of geometric, textural, and shape based features for the analysis of facial beauty. This algorithm initially develops an overall landmark model of the entire human face. A significant advantage of this methodology is the accurate model of the human face which synthesizes the geometric, textural and shape-related aspects of the face. The landmark model is then used for extracting critical characteristics which are then used in a feature vector for training using machine learning. The features extracted help to represent facial characteristics in three major areas. Geometric features help to represent the symmetrical properties and ratio-based properties of landmarks on the face. Textural features extracted help capture information related to skin texture and composition. Finally, face shape and outline features help to categorize the overall shape of a given face, which helps to represent the given female face shape and outline for further analysis of any deviations from the basic face shapes. These features are then used in a classifier to appropriately categorize each image. The database used for the source of images contains images of female subjects from a variety of backgrounds and levels of attractiveness.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.008 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".