Sharpening from Shadows: Sensor Transforms for Removing Shadows using a Single Image
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.
Bibliographic record
Abstract
Illumination conditions in images, such as shadows, can cause problems for both humans and computers. As well as shadows obscuring some features in images for human observers, many computer vision algorithms such as tracking, segmentation, recognition, and categorization are challenged by varying illumination. Previously, shadow removal algorithms were proposed that require recording a sequence of calibration images of a fixed scene over different illumination conditions, say over a day. As another alternative, calibration is replaced by using information in the single image itself, seeking a projection that minimizes entropy and allows one to generate a grayscale image that has shadows effectively eliminated. In this paper we wish to improve the entropy-based method by carrying out a sensor sharpening matrix transform first. In preceding work such a sensor transform for shadow removal was sought by utilizing many calibration images. Here, instead, we replace the calibration information by user interaction: we ask the user to identify two (or more) regions in a single image that correspond to the same surface(s) in shadow and not in shadow. Then using image data from these regions only, we generate a sensor sharpening transform via an optimization aimed at minimizing the difference between in-shadow and out-of-shadow pixel values once they are projected to grayscale. Again, entropy minimization is the driving force leading to a correct sensor matrix transform. Results show that, compared to using the camera sensors as-is, the sensor sharpening is beneficial for better shadow removal.
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 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 | 0.000 | 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 | 0.001 | 0.002 |
| 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 it