Related Experiment Video
Updated: Jun 5, 2026
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
Auto-labeling for single-photon LiDAR semantic understanding under varying acquisition conditions
Ziting Wen1, Zili Zhang2, Kemi Ding1
1School of Automation and Intelligent Manufacturing, Southern University of Science and Technology, Shenzhen, 518055, China.
Summary
Kernel-Weighted Auto-Labeling (KWAL) improves auto-labeling by considering measurement quality. This condition-aware framework reduces labeling errors in low-quality data, enhancing downstream machine learning model performance.
Area of Science:
- Computer Science
- Machine Learning
- Data Science
Background:
- Auto-labeling expands datasets using model predictions but struggles with varying measurement quality.
- Global confidence thresholds cause bias, accumulating errors in low-quality data and degrading model performance.
Purpose of the Study:
- Introduce Kernel-Weighted Auto-Labeling (KWAL), a novel framework for condition-aware auto-labeling.
- Address the challenge of varying prediction accuracy across different measurement quality ranges.
Main Methods:
- KWAL calibrates confidence and selects pseudo-labels using per-sample acquisition statistics.
- Employs a lightweight calibration model, kernel-weighted objective, and anchor-wise conditional value-at-risk loss.
- Utilizes historical prediction aggregation for decision stability across iterations.
Main Results:
- KWAL achieves high-coverage auto-labeling with consistent accuracy across measurement quality ranges.
- Significantly reduces the accuracy gap between low- and high-quality data conditions (by over 50%).
- Demonstrates effectiveness in depth-image classification and point cloud semantic segmentation tasks.
Conclusions:
- KWAL offers a robust solution for auto-labeling by accounting for sample-wise measurement quality.
- Enables more equitable performance across diverse data quality levels, improving overall model robustness.
Related Concept Videos
01:16Confocal Fluorescence Microscopy
Confocal microscopy is an advanced microscopic technique. The prime advantage of the confocal microscope over other microscopy techniques is its ability to block the out-of-focus light from the illuminated samples using pinholes. It is widely used with fluorescence optics to obtain high-resolution, sharp contrast images. Unlike optical microscopes, confocal microscopes use a focused beam of light laser to scan the entire sample surface at different z-planes. These microscopes are, therefore,...
02:16Light Acquisition
In order to produce glucose, plants need to capture sufficient light energy. Many modern plants have evolved leaves specialized for light acquisition. Leaves can be only millimeters in width or tens of meters wide, depending on the environment. Due to competition for sunlight, evolution has driven the evolution of increasingly larger leaves and taller plants, to avoid shading by their neighbors with contaminant elaboration of root architecture and mechanisms to transport water and nutrients.