MegaDescriptor-L-384 Open-source Image Feature Model - Facilitating Animal Re-identification and Serving Ecological Applications

Megadescriptor L 384

Developed by BVRA

An image feature model based on the Swin-L architecture, specifically designed for animal re-identification tasks, with wide applications in the field of ecology.

Image Classification

PyTorch

#Animal Re-identification #Swin-L Architecture #Ecological Monitoring

Downloads 5,957

Release Time : 9/27/2023

Model Overview

This model is a visual feature extraction model based on the Swin Transformer architecture, primarily used for animal re-identification tasks. It has been pre-trained on multiple wildlife datasets and can generate high-quality image embedding features.

Model Features

High-performance Feature Extraction

Based on the Swin-L architecture, capable of extracting high-quality image feature representations

Optimized for Animal Re-identification

Specifically optimized and pre-trained for animal re-identification tasks

Large Input Size Support

Supports high-resolution image inputs of 384x384 pixels

Model Capabilities

Image Feature Extraction

Animal Individual Identification

Wildlife Monitoring

Use Cases

Ecological Conservation

Wildlife Population Monitoring

Used to identify and track specific wild animal individuals, monitoring population numbers and activity ranges

Improves the efficiency and accuracy of wildlife conservation efforts

Scientific Research

Animal Behavior Research

Helps researchers identify and track specific animal individuals to study their behavioral patterns

Provides technical support for animal behavior studies

🚀 Model card for MegaDescriptor-L-384

A Swin-L image feature model superwisely pre-trained on animal re-identification datasets, aiming to solve the problem of animal re-identification and provide high - quality image feature extraction.

🚀 Quick Start

The MegaDescriptor - L - 384 is a powerful model for animal re - identification. You can quickly start using it by following the usage examples below.

✨ Features

It is a Swin - L image feature model, specifically pre - trained on animal re - identification datasets.
Capable of generating high - quality image embeddings for animal re - identification tasks.

📦 Installation

The installation details are not provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

import timm
import torch
import torchvision.transforms as T

from PIL import Image
from urllib.request import urlopen

model = timm.create_model("hf-hub:BVRA/MegaDescriptor-L-384", pretrained=True)
model = model.eval()

train_transforms = T.Compose([T.Resize(size=(384, 384)),
                              T.ToTensor(), 
                              T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])]) 

img = Image.open(urlopen(
    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
))

output = model(train_transforms(img).unsqueeze(0))  # output is (batch_size, num_features) shaped tensor
# output is a (1, num_features) shaped tensor

📚 Documentation

Model Details

Property	Details
Model Type	Animal re - identification / feature backbone
Model Stats	Params (M): 228.8 Image size: 384 x 384 Architecture: swin_large_patch4_window12_384
Paper	[WildlifeDatasets_An_Open - Source_Toolkit_for_Animal_Re - Identification](https://openaccess.thecvf.com/content/WACV2024/html/Cermak_WildlifeDatasets_An_Open - Source_Toolkit_for_Animal_Re - Identification_WACV_2024_paper.html)
Related Papers	Swin Transformer: Hierarchical Vision Transformer using Shifted Windows DINOv2: Learning Robust Visual Features without Supervision
Pretrain Dataset	All available re - identification datasets --> https://github.com/WildlifeDatasets/wildlife - datasets

📄 License

This project is licensed under the CC - BY - NC - 4.0 license.

📚 Citation

@inproceedings{vcermak2024wildlifedatasets,
  title={WildlifeDatasets: An open - source toolkit for animal re - identification},
  author={{\v{C}}erm{\'a}k, Vojt{\v{e}}ch and Picek, Lukas and Adam, Luk{\'a}{\v{s}} and Papafitsoros, Kostas},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={5953--5963},
  year={2024}
}

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