Carmodel Open-Source Vision Model - Finetuned Based on Google Model, Accurately Identified with Ultra-High F1 Score

Carmodel

Developed by TechRoC123

A vision model fine-tuned based on google/vit-base-patch16-224, achieving an F1 score of 0.9931 on the evaluation set

Downloads 24

Release Time : 12/3/2023

Model Overview

This model is an image classification model based on the Vision Transformer architecture, primarily used for vehicle-related image recognition tasks

High-precision classification

Achieves an excellent F1 score of 0.9931 on the evaluation set

Based on ViT architecture

Utilizes the Vision Transformer architecture with powerful image feature extraction capabilities

Lightweight fine-tuning

Fine-tuned on the base model with high training efficiency

Vehicle image classification

Visual feature extraction

Intelligent transportation

Vehicle type recognition

Identify vehicle types in images

F1 score 0.9931

Training Loss	Epoch	Step	Validation Loss	F1
0.1732	0.31	500	0.6651	0.8403
0.3989	0.62	1000	0.2942	0.9167
0.2136	0.93	1500	0.1782	0.9542
0.0549	1.23	2000	0.2001	0.9639
0.0287	1.54	2500	0.1304	0.9819
0.0091	1.85	3000	0.1112	0.9819
0.0039	2.16	3500	0.0667	0.9917
0.0023	2.47	4000	0.0708	0.9903
0.0002	2.78	4500	0.0635	0.9931
0.0002	3.09	5000	0.0619	0.9931
0.0002	3.4	5500	0.0730	0.9917
0.0	3.7	6000	0.0684	0.9917
0.0009	4.01	6500	0.0696	0.9917
0.0	4.32	7000	0.0693	0.9917
0.0	4.63	7500	0.0686	0.9931
0.0004	4.94	8000	0.0675	0.9931

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base