AIbase
EN
Home
Videomae Base Finetuned Kinetics Violence Nonviolence Tuned
V

Videomae Base Finetuned Kinetics Violence Nonviolence Tuned

Developed by cliffer1
A video classification model based on the VideoMAE architecture, specifically fine-tuned for violence and non-violence scene classification tasks
Video Processing
Transformers
#Video violence detection#High-precision action recognition#Fine-tuning pre-trained models
Downloads 56
Release Time : 3/10/2025

Model Overview

This model is a fine-tuned version of MCG-NJU/videomae-base-finetuned-kinetics, focusing on classifying violent and non-violent scenes in video content, achieving 98.05% accuracy on the evaluation set

Model Features

High accuracy
Achieves 98.05% accuracy in violence/non-violence classification tasks
Based on VideoMAE architecture
Utilizes efficient video masked autoencoder pre-training architecture
Fine-tuning optimization
Targeted fine-tuning based on kinetics dataset pre-training

Model Capabilities

Video content analysis
Violence scene detection
Video classification

Use Cases

Content moderation
Video platform content filtering
Automatically identifies and flags videos that may contain violent content
98.05% accuracy
Security monitoring
Surveillance video analysis
Real-time detection of violent behavior in surveillance videos
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase