无人机图像暴力行为检测方法概述

1. Conv + LSTM：Efficient Two-Stream Network for Violence Detection Using Separable Convolutional LSTM：

相邻两帧图像会做一个相减操作，

 

2.两种图像融合技术： Multi-Level Two-Stream Fusion-Based Spatio-Temporal Attention Model for Violence Detection and Localization。

 

 3. 人体关键节点检测：个人感觉不太适用于无人机图像，

4. 引入transformer结构：TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios