# Lightweight Vision Model

Vision encoder based on Mistral Small model, supports image-text generation tasks, compatible with llama.cpp framework

LSNet is a family of lightweight vision models inspired by the dynamic multi-scale capabilities of the human visual system, achieving a balance between performance and efficiency across various vision tasks.

SAM2 weights (HieraDet image encoder only) based on the timm library, derived from Facebook's Hiera small model.

Fine-tuned version of Google's ViT model for cat emotion classification

A multi-class image classification model trained using AutoTrain, capable of classifying common objects

An image classification model based on the Swin Transformer Tiny architecture, fine-tuned on the CIFAR10 dataset with an accuracy of 97.24%

A multi-class image classification model trained using AutoTrain, capable of recognizing and classifying various common objects

This is a multi-class image classification model trained using AutoTrain, capable of recognizing common objects such as tigers, teapots, and palaces.

Image classification model fine-tuned on the CIFAR10 dataset based on Swin Transformer Tiny architecture

An image classification model fine-tuned based on Google's Vision Transformer (ViT), trained on custom image datasets

Image classification model based on Swin Transformer Tiny architecture, fine-tuned for 130 epochs on a specific image dataset

This model is a fine-tuned version of facebook/levit-192 on an unlabeled dataset, demonstrating excellent performance in precision, recall, F1 score, and accuracy.

A dog food image classification model fine-tuned based on Swin Transformer Tiny architecture, achieving 98.8% accuracy on the test set

A lightweight image classification model based on Microsoft's Swin Transformer Tiny architecture, achieving 92.86% test accuracy after fine-tuning on a snack classification dataset