Related Experiment Video
Updated: Jun 28, 2026
13:19
Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
9.2K
Approximating the uncertainty of deep learning reconstruction predictions in single-pixel imaging.
Ruibo Shang1,2, Mikaela A O'Brien1, Fei Wang3,4
1Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA.
Communications Engineering
|March 11, 2024
Summary
This study introduces a Bayesian convolutional neural network (BCNN) for single-pixel imaging (SPI). The BCNN quantifies prediction uncertainty, aiding in assessing image quality and guiding system adjustments for improved deep learning reconstruction.
Area of Science:
- Optics and photonics
- Computational imaging
- Artificial intelligence
Background:
- Single-pixel imaging (SPI) offers high-speed, broad-wavelength acquisition and compact systems.
- Deep learning (DL) methods enhance image reconstruction quality in SPI compared to traditional approaches.
- Quantifying uncertainty in DL predictions is crucial for reliable image reconstruction.
Purpose of the Study:
- To develop a Bayesian convolutional neural network (BCNN) for estimating prediction uncertainty in single-pixel imaging.
- To correlate BCNN-derived uncertainty with reconstruction errors in SPI.
- To provide a tool for assessing DL model and dataset quality in practical SPI applications.
Main Methods:
- Implementation of a Bayesian convolutional neural network (BCNN) architecture.
- Training the BCNN on SPI data to predict probability distributions for each pixel.
- Evaluating the correlation between predicted uncertainty and actual reconstruction errors.
Main Results:
- The BCNN provides pixel-wise probability distributions, indicating prediction uncertainty.
- BCNN uncertainty predictions demonstrate a correlation with SPI reconstruction errors.
- Predicted uncertainty levels can guide necessary adjustments in system, data, or network parameters.
Conclusions:
- The proposed BCNN effectively quantifies uncertainty in DL-based SPI.
- This uncertainty estimation serves as a reliable indicator of prediction confidence.
- The BCNN facilitates quality assessment of DL models and datasets in SPI applications.
Related Concept Videos
00:59Propagation of Uncertainty from Random Error
An experiment often consists of more than a single step. In this case, measurements at each step give rise to uncertainty. Because the measurements occur in successive steps, the uncertainty in one step necessarily contributes to that in the subsequent step. As we perform statistical analysis on these types of experiments, we must learn to account for the propagation of uncertainty from one step to the next. The propagation of uncertainty depends on the type of arithmetic operation performed on...
01:29Application of Linearization and Approximation
A drone flying through complex terrain often relies on more than one sensing method to estimate small changes in altitude. Along with direct measurements, air pressure provides a useful indirect indicator of vertical movement. Atmospheric pressure decreases as altitude increases, and this relationship is commonly described using an exponential model. Although accurate, converting pressure measurements into altitude values requires calculations that are too complex to perform repeatedly during...