File size: 2,807 Bytes
9729764 |
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 |
---
datasets:
- garythung/trashnet
metrics:
- accuracy
- precision
- recall
pipeline_tag: image-classification
---
### **Model Card: Trash Classification Using MobileNetV2**
---
## **Model Details**
- **Model Name**: Trash Classification CNN with MobileNetV2
- **Model Type**: Convolutional Neural Network (CNN)
- **Architecture**: MobileNetV2
- **Dataset**: [TrashNet Dataset](https://huggingface.co/datasets/garythung/trashnet)
- **Languages**: None (Image-based model)
- **License**: MIT
---
## **Model Description**
This model classifies images of trash into six categories:
- **trash**
- **plastic**
- **cardboard**
- **metal**
- **paper**
- **glass**
The model is designed to assist in waste segregation and recycling initiatives by automating the identification of waste types. It uses MobileNetV2, a lightweight CNN architecture pre-trained on ImageNet, fine-tuned on the TrashNet dataset for this specific task.
---
## **Intended Use**
### **Primary Use Cases**
- Waste management systems to automate sorting.
- Educational tools for teaching about recycling and waste segregation.
- Integration into mobile or web applications for real-time waste classification.
### **Limitations**
- Model performance may degrade with images of poor quality or those significantly different from the training dataset.
- Currently supports only six predefined trash categories.
---
## **Performance Metrics**
- **Training Accuracy**: 95%
- **Testing Accuracy**: 90%
- **Metrics Evaluated**: Accuracy, Precision, Recall, F1-score
- **Confusion Matrix**: [Available in evaluation results]
---
## **How to Use the Model**
### **Input Format**
- Images resized to 224x224 pixels and normalized to a range of 0-1.
### **Output**
- A probability distribution over six classes with the predicted label.
### **Code Example**
```python
from transformers import pipeline
from PIL import Image
# Load pre-trained model
classifier = pipeline("image-classification", model="your-model-id")
# Load an image
image = Image.open("sample_image.jpg")
# Perform classification
results = classifier(image)
print(results)
```
---
## **Training Details**
- **Framework**: TensorFlow/Keras
- **Optimizer**: Adam
- **Learning Rate**: 0.001
- **Loss Function**: Categorical Crossentropy
- **Batch Size**: 32
- **Epochs**: 20
### **Data Preprocessing**
- Images were resized to 224x224 pixels and normalized.
- Oversampling and data augmentation techniques (rotation, zoom, and rescaling) were applied to handle class imbalance and enhance generalization.
---
This model card is designed to comply with Hugging Face standards and can be adapted further as needed. Let me know if you need any specific sections expanded! |