Any ideas about Hugging Face Models For Plant disease detection

[vishmividanege]

Select Topic Area

Question

Body

Suggest Me some With your Experiences

[Tharunchndrn]

Hugging Face provides several powerful pretrained models that can be used for plant disease detection through image classification. These models are typically based on deep learning architectures such as Convolutional Neural Networks (CNNs) and Vision Transformers (ViT).

Vision Transformer (ViT) – e.g., google/vit-base-patch16-224.
This model divides images into patches and processes them using transformer architecture. It achieves high accuracy and is suitable for fine-tuning on plant disease datasets such as PlantVillage.

ResNet (Residual Network) – e.g., microsoft/resnet-50.
ResNet is a CNN-based model that uses residual connections to improve deep network training. It is computationally efficient and works well for leaf disease classification tasks.

MobileNet – e.g., google/mobilenet_v2_1.0_224.
MobileNet is a lightweight model optimized for mobile and edge devices, making it suitable for real-time plant disease detection applications for farmers.

[Sameer-710]

Yes, there are several good options on Hugging Face for plant disease detection, especially for image classification tasks.
If you're just getting started, I’d recommend looking for models already fine-tuned on the PlantVillage dataset on Hugging Face — these will save you a lot of training time.

In general:
ViT (Vision Transformer) models (like google/vit-base-patch16-224) work very well if you have enough data and compute.
ResNet-50 is a very solid and reliable choice for plant leaf disease classification.
MobileNetV2 is better if you plan to deploy on mobile or edge devices (lightweight and fast).

If your goal is:
Research / high accuracy → go with ViT or ResNet.
Mobile / real-time detection → go with MobileNet.
Fast prototype → use a model already fine-tuned on PlantVillage from the Hugging Face Hub.

[rishu072]

This project uses pretrained vision models from Hugging Face such as Vision Transformers (ViT), ConvNeXt, and Swin Transformer to classify plant leaf images into healthy or diseased categories.
The system also integrates Explainable AI (XAI) techniques like Grad-CAM to highlight the infected regions on the leaf, increasing trust and interpretability.

A web-based interface (Gradio/Streamlit) allows users to upload leaf images and receive instant predictions.