EMO-MoviNet:通过EvoNorm,Mish激活和最佳选择来提高视频中的动作识别,以实现高效的移动部署
Tarique Hussain1, Zulfiqar Ali Memon1, Rizwan Qureshi1
1Fast School of Computing, National University of Computer and Emerging Sciences, Karachi Campus, Karachi 75030, Pakistan.
Sensors (Basel, Switzerland)
|October 14, 2023
概括
本研究介绍了EMO-MoviNet-A2*,一个用于视频动作识别的深度神经网络. 它通过使用Evolving Normalization和最佳选择实现了更高的准确性和效率,大大降低了计算成本.
科学领域:
- 计算机视觉 计算机视觉
- 深度学习 (Deep Learning) 是一种深度学习.
- 人工智能的人工智能
背景情况:
- 视频中的动作识别是计算密集的.
- 像MoviNet这样的现有模型平衡了准确性和效率.
- 概括差距和关键选择是视频分析中的挑战.
研究的目的:
- 开发一个深度神经网络,以提高动作识别精度和降低计算成本.
- 提出一个经过修改的EMO-MoviNet-A2*架构,集成EvoNorm,Mish激活和最佳选择.
- 解决概括差距,改善视频动作识别中的关键选择.
主要方法:
- 在UCF101和HMDB51数据集上实施了经过修改的EMO-MoviNet-A2*架构.
- 将批量规范化取代为演进规范化 (EvoNorm),以统一规范化和激活.
- 开发了最佳框架选择 (OFS) 以有效识别关键框架.
主要成果:
- 在UCF101 (20类) 上,EMO-MoviNet-A2* 实现了0.8-1%的精度改进,其中包括OFS和Mish.
- 该模型在UCF101.1.上使用的FLOP减少了86%,参数减少了90%.
- 与MARS相比,HMDB51的FLOP和参数显著减少,在MARS51上获得了5-7%的更高准确性.
结论:
- 拟议的EMO-MoviNet-A2*模型提高了动作识别的效率和准确性.
- EvoNorm和OFS有效地解决了概括差距,并改善了关键框架选择.
- 这种架构可显著减少用于视频分析的计算资源.
相关概念视频
01:06Fixed Action Patterns
16.0K
A fixed action pattern (FAP) is a specific, hard-wired sequence of behaviors that occurs in response to an external stimulus, called a sign stimulus. The behavior is “fixed” because it is essentially unchangeable—proceeding similarly across individuals of a species every time it occurs.
16.0K
01:24Muscle Coordination and Action
1.5K
Muscle coordination is a complex and finely tuned process essential for smooth and purposeful movements like flexion, extension, adduction, abduction, and rotation. The human body orchestrates the actions of various muscles working in concert, each with a specific role. Four functional types describe how muscles work together: agonist, antagonist, synergist, and fixator.
Agonists
Agonist muscles, often called prime movers, are the primary muscles responsible for producing a specific movement....
Agonists
Agonist muscles, often called prime movers, are the primary muscles responsible for producing a specific movement....
1.5K
01:22Force Classification
1.2K
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,...
1.2K