Bibliographic record
Abstract
In this thesis, a novel shape-based representation and description method is introduced. In the proposed technique, an image is vectorized, and the resultant vector is smoothed and sampled. The vectorization step is performed by using the Hilbert space-filling curve which is known for its locality preservation. The smoothing step is conducted by using wavelet approximation. The smoothed and sampled vector, called the Shape Feature Vector (SFV), provides a concise representation of the object that is smaller than the original image by orders of magnitude. The proposed representation is not only invariant to translation, scaling, and stretching, but also robust to visual transformations such as occlusion and articulation. Distance measures that use a linear mapping between the SFV elements do not always reflect the similarity between one SFV and another. This is particularly noticeable with similar SFVs, where some parts of one SFV are slightly shifted in either direction. To help remedy this situation, the Minimum Landscape Distance (MLD) measure is proposed. MLD finds a non-linear mapping between the two sequences that reflects their structural similarities. The newly developed representation and description method is enhanced in two significant ways. First, the Dynamic Hilbert Curve (DHC) is introduced to increase the spectrum of the visual transformations that the new technique can handle. Secondly, when the class of each image database object is known, this knowledge is used to identify the Key Feature Points (KFPs) of the class. At search time, when a search object is compared with a database object, only the SFV elements that correspond to the KFPs are compared. Both improvements result in a significant increase in the technique's retrieval accuracy. Two specific application areas that benefit from the characteristics of the proposed representation are identified. First, the proposed representation is proven to be useful in providing an efficient and effective representation of two types of 3D object representations. Secondly, the newly devised representation is adopted as the basis of another representation that allows the retrieval of images based on the verbal shape descriptions of one or more of image's parts. In both cases, experimental evidence is generated to support the claims made.
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 distilled prediction
Teacher imitationNot 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.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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.
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 teacher head, 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".