Model Card for culico-net-cls-v1

culico-net-cls-v1 - image classification model focused on identifying mosquito species. This model is a result of the CulicidaeLab project and was developed by fine-tuning the tiny_vit_21m_224.dist_in22k_ft_in1k model.

The culico-net-cls-v1 is a TinyViT image classification model. It was pretrained on the large-scale ImageNet-22k dataset using distillation and then fine-tuned on the ImageNet-1k dataset by the original paper's authors. This foundational training has been further adapted for the specific task of mosquito species classification using a dedicated dataset.

Model Details:

• Model Type: Image classification / feature backbone
• Model Stats:
  • Parameters (M): 21.2
  • GMACs: 4.1
  • Activations (M): 15.9
  • Image size: 224 x 224
• Papers:
  • TinyViT: Fast Pretraining Distillation for Small Vision Transformers: https://arxiv.org/abs/2207.10666
  • Original GitHub Repository: https://github.com/microsoft/Cream/tree/main/TinyViT
• Dataset: The model was trained on the iloncka/mosquito-species-classification-dataset. This is one of a suite of datasets which also includes iloncka/mosquito-species-detection-dataset and iloncka/mosquito-species-segmentation-dataset. These datasets contain images of various mosquito species, crucial for training accurate identification models. For instance, some datasets include species like Aedes aegypti, Aedes albopictus, and Culex quinquefasciatus, and are annotated for features like normal or smashed conditions.
• Pretrain Dataset: ImageNet-22k, ImageNet-1k

Model Usage:

The model can be used for image classification tasks. Below is a code snippet demonstrating how to use the model with the Fastai library:

from fastai.vision.all import load_learner
from PIL import Image

# It is assumed that the model has been downloaded locally
learner = load_learner(model_path)
_, _, probabilities = learner.predict(image)

The CulicidaeLab Project:

The culico-net-cls-v1 model is a component of the larger CulicidaeLab project. This project aims to provide a comprehensive suite of tools for mosquito monitoring and research. Other parts of the project include:

• Datasets:
  • iloncka/mosquito-species-detection-dataset
  • iloncka/mosquito-species-segmentation-dataset
  • iloncka/mosquito-species-classification-dataset
• Python Library: https://github.com/iloncka-ds/culicidaelab
• Mobile Applications:
• • https://gitlab.com/mosquitoscan/mosquitoscan-app
  • https://github.com/iloncka-ds/culicidaelab-mobile
• Web Application: https://github.com/iloncka-ds/culicidaelab-server

Practical Applications:

The culico-net-cls-v1 model and the broader CulicidaeLab project have several practical applications:

• Integration into Third-Party Products: The models can be integrated into existing applications for plant and animal identification to expand their functionality to include mosquito recognition.
• Embedded Systems (Edge AI): These models can be optimized for deployment on edge devices such as smart traps, drones, or cameras for in-field monitoring without requiring a constant internet connection.
• Accelerating Development: The pre-trained models can serve as a foundation for transfer learning, enabling researchers to develop systems for identifying other insects or specific mosquito subspecies more efficiently.
• Expert Systems: The model can be used as a "second opinion" tool to assist specialists in quickly verifying species identification.

Acknowledgments:

The development of CulicidaeLab is supported by a grant from the Foundation for Assistance to Small Innovative Enterprises (FASIE).