A 19k-parameter pixel-wise exposure correction model, trained by treating multi-exposure copies of a photo as each other's ground truth, matches larger supervised models on a new radiometric dataset and improves edge detection, despite mixed results on the standard MSEC benchmark.
In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Unsupervised Exposure Correction
A 19k-parameter pixel-wise exposure correction model, trained by treating multi-exposure copies of a photo as each other's ground truth, matches larger supervised models on a new radiometric dataset and improves edge detection, despite mixed results on the standard MSEC benchmark.