MobileCLIP-S2-OpenCLIP Open-source Image-text Model - A Practical Choice for Fast Zero-shot Image Classification

Mobileclip S2 OpenCLIP

Developed by apple

MobileCLIP-S2 is an efficient text-image model that achieves fast zero-shot image classification through multimodal reinforcement training.

Text-to-Image

Safetensors

#Zero-shot image classification #Low-latency inference #Multimodal reinforcement training

Downloads 99.74k

Release Time : 6/7/2024

Model Overview

MobileCLIP-S2 is a medium-sized variant in the MobileCLIP series, specifically designed for fast zero-shot image classification tasks, delivering excellent classification performance while maintaining efficient inference speed.

Model Features

Efficient performance

Outperforms SigLIP's ViT-B/16 model in zero-shot performance, 2.3x faster and 2.1x smaller in size

Low training data requirement

Uses only 13B training samples, 3x fewer than similar models

Multimodal reinforcement training

Employs specialized multimodal training methods to enhance model performance

Model Capabilities

Zero-shot image classification

Text-image matching

Multimodal understanding

Use Cases

Computer vision

Image classification

Classifies images without specific training

Achieves 74.4% zero-shot Top-1 accuracy on ImageNet-1k

Visual search

Searches for relevant images based on text descriptions

Mobile applications

Mobile image recognition

Enables efficient image recognition on mobile devices

Low latency (3.6ms for image + 3.3ms for text)

Property	Details
Model Type	[MobileCLIP - S0](https://hf.co/pcuenq/MobileCLIP - S0), [MobileCLIP - S1](https://hf.co/pcuenq/MobileCLIP - S1), [MobileCLIP - S2](https://hf.co/pcuenq/MobileCLIP - S2), [MobileCLIP - B](https://hf.co/pcuenq/MobileCLIP - B), [MobileCLIP - B (LT)](https://hf.co/pcuenq/MobileCLIP - B - LT)
# Seen Samples (B)	13 (for S0, S1, S2, B); 36 (for B (LT))
# Params (M) (img + txt)	11.4 + 42.4 (S0); 21.5 + 63.4 (S1); 35.7 + 63.4 (S2); 86.3 + 63.4 (B, B (LT))
Latency (ms) (img + txt)	1.5 + 1.6 (S0); 2.5 + 3.3 (S1); 3.6 + 3.3 (S2); 10.4 + 3.3 (B, B (LT))
IN - 1k Zero - Shot Top - 1 Acc. (%)	67.8 (S0); 72.6 (S1); 74.4 (S2); 76.8 (B); 77.2 (B (LT))
Avg. Perf. (%) on 38 datasets	58.1 (S0); 61.3 (S1); 63.7 (S2); 65.2 (B); 65.8 (B (LT))

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

Mobileclip S2 OpenCLIP

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 MobileCLIP: Fast Image-Text Models through Multi-Modal Reinforced Training

🚀 Quick Start

✨ Features

📚 Documentation

Checkpoints

📄 License