Videomae-base-finetuned-kinetics Open-source Video Understanding Model - Accuracy on Evaluation Set Reaches 91.23%

Videomae Base Finetuned Kinetics 0409 Final 5sec Org Ab7 Val As123 Retry

Developed by d2o2ji

A video understanding model fine-tuned based on MCG-NJU/videomae-base-finetuned-kinetics, achieving 91.23% accuracy on the evaluation set

Downloads 30

Release Time : 4/9/2025

Model Overview

This model is a video classification model based on the VideoMAE architecture, specifically optimized for action recognition tasks.

High Accuracy

Achieves 91.23% accuracy on the evaluation set

Based on VideoMAE Architecture

Utilizes video masked autoencoder pre-training framework

5-second Video Processing

Specially optimized for processing 5-second video clips

Video Action Recognition

Video Classification

Temporal Behavior Understanding

Video Analysis

Action Recognition

Recognize human actions and behaviors in videos

91.23% accuracy

Video Content Classification

Automatically classify video clips

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6536	0.0100	659	0.3131	0.8889
0.0064	1.0100	1318	0.2702	0.9241
0.0189	2.0100	1977	0.6265	0.8271
0.0018	3.0100	2636	0.5558	0.8709
0.0105	4.0100	3295	0.8117	0.8443
0.0026	5.0100	3954	0.4269	0.9014
0.0107	6.0100	4613	0.5239	0.8638
0.0481	7.0100	5272	0.4894	0.8740
0.0069	8.0100	5931	0.5610	0.8599
0.0035	9.0100	6590	0.5693	0.8795
1.3138	10.0100	7249	1.0626	0.8020
0.0002	11.0100	7908	0.4887	0.8897

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base