Open-source model_timesformer_subset_02 video understanding model - Accuracy reaches 88.52% after fine-tuning on an unknown dataset

Model Timesformer Subset 02

Developed by namnh2002

A video understanding model based on the TimeSformer architecture, fine-tuned on an unknown dataset with an accuracy of 88.52%

Video Processing

Transformers

#Video Understanding #High Accuracy #Temporal Modeling

Downloads 15

Release Time : 3/4/2024

Model Overview

This is a video classification model based on the TimeSformer architecture, suitable for temporal video understanding tasks. The model performs exceptionally well on the evaluation set, achieving an accuracy of 88.52%.

Model Features

High Accuracy

Achieves an accuracy of 88.52% on the evaluation set, demonstrating excellent performance

Temporal Understanding

Based on the TimeSformer architecture, excels at processing temporal information in videos

Efficient Training

Utilizes linear learning rate scheduling and Adam optimizer, ensuring stable and efficient training

Model Capabilities

Video Classification

Temporal Feature Extraction

Video Content Understanding

Use Cases

Video Analysis

Action Recognition

Recognize human actions or behaviors in videos

Accuracy: 88.52%

Scene Classification

Classify the scene of video content

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.9368	0.1	625	1.8302	0.5026
0.9936	1.1	1250	1.3368	0.6081
0.9407	2.1	1875	1.1348	0.6794
0.8338	3.1	2500	0.9604	0.7270
0.629	4.1	3125	0.7775	0.7684
0.4094	5.1	3750	0.6939	0.8056
0.398	6.1	4375	0.5883	0.8366
0.3242	7.1	5000	0.4594	0.8707
0.2768	8.1	5625	0.5158	0.8604
0.2571	9.1	6250	0.4130	0.8852

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご

Model Timesformer Subset 02

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 model_timesformer_subset_02

🚀 Quick Start

📚 Documentation

Model description

Intended uses & limitations

Training and evaluation data

🔧 Technical Details

Training procedure

Training hyperparameters

Training results

Framework versions

📄 License