File size: 3,937 Bytes
a9dce72 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 |
# 🧠 Beans-Image-Classification-AI-Model
A fine-tuned image classification model trained on the Beans dataset with 3 classes: angular_leaf_spot, bean_rust, and healthy. This model is built using Hugging Face Transformers and the ViT (Vision Transformer) architecture and is suitable for educational use, plant disease classification tasks, and image classification experiments.
---
## ✨ Model Highlights
- 📌 Base Model: google/vit-base-patch16-224-in21k
- 📚 Fine-tuned: Beans dataset
- 🌿 Classes: angular_leaf_spot, bean_rust, healthy
- 🔧 Framework: Hugging Face Transformers + PyTorch
- 📦 Preprocessing: AutoImageProcessor from Transformers
---
## 🧠 Intended Uses
- ✅ Educational tools for training and evaluation in agriculture and plant disease detection
- ✅ Benchmarking vision transformer models on small datasets
- ✅ Demonstration of fine-tuning workflows with Hugging Face
---
## 🚫 Limitations
- ❌ Not suitable for real-world diagnosis in agriculture without further domain validation
- ❌ Not robust to significant background noise or occlusion in images
- ❌ Trained on small dataset, may not generalize beyond bean leaf diseases
---
📏 Input & Output
- Input: RGB image of a bean leaf (expected size 224x224)
- Output: Predicted class label — angular_leaf_spot, bean_rust, or healthy
---
## 🏋️♂️ Training Details
| Attribute | Value |
|--------------------|----------------------------------|
| Base Model |`google/vit-base-patch16-224-in21k|
| Dataset |Beans Dataset (train/val/test) |
| Task Type | Image Classification |
| Image Size | 224 × 224 |
| Epochs | 3 |
| Batch Size | 16 |
| Optimizer | AdamW |
| Loss Function | CrossEntropyLoss |
| Framework | PyTorch + Transformers |
| Hardware | CUDA-enabled GPU |
---
## 📊 Evaluation Metrics
| Metric | Score |
| ----------------------------------------------- | ----- |
| Accuracy | 0.98 |
| F1-Score | 0.99 |
| Precision | 0.98 |
| Recall | 0.99 |
---
---
🚀 Usage
```python
from transformers import AutoImageProcessor, AutoModelForImageClassification
from PIL import Image
import torch
model_name = "AventIQ-AI/Beans-Image-Classification-AI-Model"
processor = AutoImageProcessor.from_pretrained(model_name)
model = AutoModelForImageClassification.from_pretrained(model_name)
model.eval()
def predict(image_path):
image = Image.open(image_path).convert("RGB")
inputs = processor(images=image, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model(**inputs)
preds = torch.argmax(outputs.logits, dim=1)
return model.config.id2label[preds.item()]
# Example
print(predict("example_leaf.jpg"))
```
---
- 🧩 Quantization
- Post-training static quantization applied using PyTorch to reduce model size and accelerate inference on edge devices.
----
🗂 Repository Structure
```
.
beans-vit-finetuned/
├── config.json ✅ Model architecture & config
├── pytorch_model.bin ✅ Model weights
├── preprocessor_config.json ✅ Image processor config
├── special_tokens_map.json ✅ (Auto-generated, not critical for ViT)
├── training_args.bin ✅ Training metadata
├── README.md ✅ Model card
```
---
🤝 Contributing
Open to improvements and feedback! Feel free to submit a pull request or open an issue if you find any bugs or want to enhance the model.
|