Image Separation using Multi-layer Image Segmentation for Translucent Partially Overlapped Objects
Bibliographic record
Abstract
One way to solve under-determined image separation is to use statistical information about the type of data to be decomposed.In this dissertation, we propose a two stage method for cervical cell separation.In the first stage, we propose a CNN-based multi-layer random walker image segmentation method.The results of image segmentation at the first stage are then used as the side information for the cervical cell separation in the second stage.Convolutional neural networks (CNNs) are recently used in computer vision applications such as image segmentation.One of the biggest advantages of CNNs is that they extract important features automatically from the data.However, CNNs usually use a high number of data for training but it is not always possible to find enough training data for some applications.Also, CNNs are good at generalizing the training, but not for finding the accurate edges of cervical cells at the current implementations for cervical cell segmentation.One solution to improve CNN segmentation results is to use a post-processing method as edge refinement.Random walker image segmentation method is a graph-based image segmentation method that is good at region growing; so, it can be used as the refinement step.However, random walker is sensitive to initial setup, so if the seeds are not extracted correctly, the final segmentation results will be poor.In this dissertation we use a CNN-based random walker image segmentation approach for cervical cell segmentation.Different from other methods that use CNN binary segmentation results for fine tuning, CNN probabilistic map is utilized to guide random walker image segmentation method at the refinement step.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".