相关实验视频
Updated: May 31, 2025
05:16
Flying Insect Detection and Classification with Inexpensive Sensors
Published on: October 15, 2014
25.1K
使用红外传感器图像的物体与未知拒绝 (SCOUR) 的同时分类
Adam Cuellar1, Daniel Brignac2, Abhijit Mahalanobis2
1Center for Research in Computer Vision, University of Central Florida, Orlando, FL 32816-8005, USA.
Sensors (Basel, Switzerland)
|January 25, 2025
概括
这项研究引入了一种新的方法,通过增强分类器来拒绝未知的物体来改善红外目标识别. 二级网络可以识别未知的目标,而不需要重新训练主要分类器,从而改善防御和安全应用.
科学领域:
- 计算机视觉 计算机视觉
- 机器学习 机器学习
- 防务技术 防务技术 技术
背景情况:
- 红外目标识别对于国防和安全至关重要.
- 现有的分类器很难在没有错误分类的情况下拒绝未知的对象.
- 需要强大的系统来识别已知的目标并拒绝新的威胁.
研究的目的:
- 增强预先训练的分类器来检测和拒绝未知的类.
- 保持已知类的分类器性能.
- 开发一种不需要OOD数据进行培训的方法.
主要方法:
- 引入了二次回归网络,以与初级分类器一起工作.
- 组合初级分类器的信心与二级网络的类条件得分.
- 利用贝叶斯框架来改善已知和未知对象的分离.
主要成果:
- 在CIFAR-10和中波红外线 (MWIR) 数据集上证明了有效性.
- 在拒绝未知的目标类型方面超过了最先进的方法.
- 保持已知目标的准确分类.
结论:
- 拟议的方法有效地增强了红外目标识别系统.
- 成功地将未知对象从未知类中分离出来,没有OOD训练数据.
- 为需要强大的目标识别的国防和安全应用提供了有前途的解决方案.
相关概念视频
01:03IR Frequency Region: Fingerprint Region
729
IR spectra are divided into two main regions: the diagnostic region and the fingerprint region. The diagnostic region of the spectrum lies above 1500 cm−1. The absorptions resulting from single-bond vibrations of the N–H, C–H, and O–H stretch at higher wavenumbers and appear on the left side of the spectrum. The stretching absorptions of the C≡C and C≡N occur between 2100–2300 cm−1. In contrast, those arising from stretching absorptions of the...
729
01:09Infrared (IR) Spectroscopy: Overview
1.4K
When electromagnetic radiation passes through a material, atoms or molecules transition from a lower to a higher energy state by absorbing radiation corresponding to the energy difference between the two states. The absorption of infrared (IR) radiation causes transitions between vibrational energy levels in a molecule. Therefore, IR spectroscopy is a useful analytical tool for determining the molecular structure of molecules.
Different compounds display unique properties due to their...
Different compounds display unique properties due to their...
1.4K