Technical note: The impact of image size on bloodstain pattern analysis using machine learning
Bibliographic record
Abstract
Classification of bloodstain patterns can occur at the scene or through (digital) images submitted for analysis and/or peer review. Capturing high-resolution images requires high-quality cameras or scanners and can be time consuming, especially when image stitching is needed. With the increasing use of machine learning (ML) for image analysis, handling large images is also computationally demanding. This raises the question of whether high-resolution images are necessary for accurate ML based pattern classification. In this study, we explore the role of image size on bloodstain classification by replicating an existing experiment distinguishing impact versus forward spatter, using both original and resized images. Despite expectations, reducing resolution did not significantly affect classification accuracy. This suggests that modern ML techniques may be less reliant on fine image detail than human analysts. We also examined the importance of feature selection in classification. Using only four key features showed results comparable to those obtained using all 58 from the original model. This points to the potential for simpler, more efficient models without sacrificing accuracy. These results raise important questions, while the model performed well under reduced input conditions, it is essential to understand what it relies on and why. Rather than applying ML models blindly, we emphasize the need to understand their behavior, for responsible and effective use in forensic workflows.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 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".