通过临时-YOLOv8实现多功能小型物体检测
Martin C van Leeuwen1, Ella P Fokkinga1, Wyke Huizinga1
1TNO, Defence, Safety and Security, 2597 AK The Hague, The Netherlands.
Sensors (Basel, Switzerland)
|November 27, 2024
概括
这项研究通过结合时间视频上下文和专门的数据增强来增强使用深度学习的小物体检测. 改进的YOLOv8模型实现了显著更高的准确性,在各种环境中展示了有效的检测.
科学领域:
- 计算机视觉 计算机视觉
- 人工智能的人工智能
- 机器学习 机器学习
背景情况:
- 精确检测小物体是使用深度学习进行自动化物体检测的持续挑战.
- 现有的深度学习探测器通常会忽略视频中的有价值的时间信息,这对于低信号对噪声情况至关重要.
- 目前用于小物体检测的数据集往往是特定于任务的,缺乏多样性,并遭受糟糕的注释.
研究的目的:
- 开发一个通用的深度学习管道,用于准确检测小物体.
- 解决当前方法的局限性,包括特征的独特性,时间信息的利用和数据集质量.
- 改进现有的物体检测架构,如YOLOv8用于小物体识别.
主要方法:
- 从视频数据中利用时间上下文来增强特征表示.
- 实施专门为小型对象设计的数据增强技术.
- 利用包括各种民用和军事物体在内的内部数据集进行模型培训和验证.
- 将性能与基线YOLOv8和在公共数据集上训练的模型进行比较.
主要成果:
- 在YOLOv8中实现了显著的性能提升,将平均精度 (mAP) 从0.465提高到0.839.
- 证明了整合时间信息和定制数据增强的有效性.
- 展示了在多样化,精心策划的数据集上训练的模型对环境特定模型的优越性.
- 验证了模型在各种环境中准确检测小物体的能力.
结论:
- 拟议的深度学习管道通过利用时间上下文和专业增强来显著提高小物体检测的准确性.
- 一个多样化且注释良好的数据集对于开发强大的小物体探测器至关重要.
- 增强的YOLOv8架构为广泛的应用中检测小物体提供了快速而准确的解决方案.
相关概念视频
01:05Difference from Background: Limit of Detection
5.9K
The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...
The LOD indicates the presence or absence...
5.9K
01:22Force Classification
1.1K
Forces play a crucial role in the study of physics and engineering. They are essential in describing the motion, behavior, and equilibrium of objects in the physical world. Forces can be classified based on their origin, type, and direction of action.
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...
1.1K
01:20Deconvolution
132
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
132
01:10Detection of Black Holes
2.2K
Although black holes were theoretically postulated in the 1920s, they remained outside the domain of observational astronomy until the 1970s.
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
2.2K
01:19Masking and Demasking Agents
2.3K
EDTA titrations may necessitate masking and demasking agents to temporarily protect a particular metal ion in a mixture from the EDTA reaction. These agents facilitate the sequential analysis of the metal ions by forming stable complexes with some—but not all—metal ions during certain steps.
There are many masking agents, such as cyanide, fluoride, triethanolamine, thiourea, and 2,3-bis(sulfanyl)propan-1-ol (formerly 2,3-dimercapto-1-propanol), with the masking agent chosen based on...
There are many masking agents, such as cyanide, fluoride, triethanolamine, thiourea, and 2,3-bis(sulfanyl)propan-1-ol (formerly 2,3-dimercapto-1-propanol), with the masking agent chosen based on...
2.3K
01:29Aggregates Classification
305
Aggregate classification is generally based on its size, petrographic characteristics, weight, and source. Size classification ranges from coarse to fine aggregates, defined by the size of the particles. Coarse aggregates are particles that do not pass through ASTM sieve No. 4, and aggregates that pass through the sieve are fine aggregates.
Petrographic classification groups aggregates based on common mineralogical characteristics. Some of the common mineral groups found in aggregates are...
Petrographic classification groups aggregates based on common mineralogical characteristics. Some of the common mineral groups found in aggregates are...
305