在视听监控数据中,基于嵌入的对生成用于对比表示学习
Wei-Cheng Wang1, Sander De Coninck1, Sam Leroux1
1IDLab, Ghent University-imec, Ghent, Belgium.
Frontiers in robotics and AI
|January 28, 2025
概括
这项研究引入了一种用于智能城市监控的新自主监督学习方法,通过生成更好的对比对来克服当前模型的局限性来改进视听数据分析.
科学领域:
- 人工智能的人工智能
- 计算机视觉 计算机视觉
- 机器学习 机器学习
背景情况:
- 智慧城市利用传感器,如麦克风和RGB摄像头,以确保公民的安全和舒适.
- 自主监督学习,特别是对比学习,对于分析视听数据至关重要,因为注释成本很高.
- 现有的视听对比学习方法面临着虚假负面和信息瓶的挑战.
研究的目的:
- 为了解决智能城市监控的视听对比学习的局限性.
- 提出一种新的方法来产生更有效的对比对.
- 为了改善下游任务的视听表现的学习.
主要方法:
- 开发了一种基于嵌入模式之间的距离来生成对比对的新方法,超越仅仅是时间线索.
- 引入了多个正数的新损失函数.
- 实验验证了对现实世界监控数据的方法.
主要成果:
- 提出的方法有效地产生语义同步的对,缓解信息瓶.
- 学习的表示适用于各种下游任务,如事件本地化,异常检测和事件搜索.
- 实现了与最先进的专业方法相美的性能.
结论:
- 这种新的对比学习方法增强了智能城市应用程序的视听表现学习.
- 该方法有效地应对了监控数据中虚假阴性和信息瓶等挑战.
- 学习的表示形式为多个关键智能城市监控任务提供了多功能基础.
相关概念视频
Sign Test for Matched Pairs
96
The sign test for matched pairs offers a robust method for comparing two paired samples, often for the effects of an intervention in one of them. This method is very useful in situations where the underlying distribution of the data is unknown. The test compares two related samples—often pre- and post-treatment measurements on the same subjects—to determine if there are significant differences in their median values.
To conduct the sign test, we first calculate the differences in...
To conduct the sign test, we first calculate the differences in...
96
Associative Learning
285
Associative learning is a fundamental concept in behavioral psychology, wherein a connection is established between two stimuli or events, leading to a learned response. This process is critical in understanding how behaviors are acquired and modified. Conditioning, the mechanism through which associations are formed, can be divided into two main types: classical conditioning and operant conditioning, each elucidating different aspects of associative learning.
Classical conditioning, also known...
Classical conditioning, also known...
285
Observational Learning
123
Albert Bandura's observational learning, also known as imitation or modeling, occurs when a person observes and imitates another's behavior. It is a quicker process than operant conditioning. A well-known example is the Bobo doll study, where children who saw an adult acting aggressively towards the doll were more likely to act aggressively when left alone, compared to those who observed a nonaggressive adult. Many psychologists view observational learning as a form of latent learning...
123
Equivalent Couples
274
In mechanical engineering, the concept of equivalent couples plays a crucial role in understanding and analyzing various mechanical systems.
Two couples are considered to be equivalent if they produce the same rotational effect on a rigid body. In other words, the two couples have the same magnitude and act in the same direction, causing the same angular displacement or acceleration in the body.
For instance, consider two couples lying in the plane of the page, with one having a pair of equal...
Two couples are considered to be equivalent if they produce the same rotational effect on a rigid body. In other words, the two couples have the same magnitude and act in the same direction, causing the same angular displacement or acceleration in the body.
For instance, consider two couples lying in the plane of the page, with one having a pair of equal...
274
Masking 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
Per-Unit Sequence Models
67
An ideal Y-Y transformer, grounded through neutral impedances, displays per-unit sequence networks akin to those of a single-phase ideal transformer when subjected to balanced positive- or negative-sequence currents. These currents do not produce neutral currents, and their associated voltage drops.
Zero-sequence currents, which are identical in magnitude and phase, generate a neutral current, resulting in voltage drops across the neutral impedance and the low-voltage winding. If the...
Zero-sequence currents, which are identical in magnitude and phase, generate a neutral current, resulting in voltage drops across the neutral impedance and the low-voltage winding. If the...
67


