Update README.md

README.md

@@ -2,7 +2,8 @@
 license: creativeml-openrail-m
 ---
 
-Based on:
+# Model Card for Kanji_ETL8G
+### Summary:
 ```
 ETL8B
  - 153916 samples
@@ -12,47 +13,39 @@ ETL8B
  - image_height: 63px
 ```
 
-I was testing a few more samples locally with the below.   Note the results of the model are encoded. 
-
-
-```
-
-from PIL import Image
-import numpy as np
+## Model Details
+- **Model Name:** Kanji_ETL8G
+- **Version:** 1.0.0
+- **Model Type:** Neural Network 
+- **Framework:** PyTorch
+  
+## Model Description
+This model is trained on a dataset derived from the ETL8G dataset to recognize Kanji characters from 64x64 grayscale images. The primary use-case is for optical character recognition (OCR) for handwritten Kanji characters.
+
+## Intended Use
+The primary application of this model is for OCR tasks to recognize handwritten Kanji characters in images, with potential extensions for applications like smart dictionary lookup, handwriting-based user authentication, and so on.
+
+## Limitations
+This model might have limitations regarding:
+- Variability in handwriting styles not present in the training set. (161 samples per character/class were used)
+- Noises and artifacts in input images.
+- Characters written in unconventional ways.
+## Data Details
+### Training Data:
+- **Dataset:** Derived from the ETL8G dataset (hhttp://etlcdb.db.aist.go.jp/specification-of-etl-8)
+- **Size:** 153916 samples
+- **Data Type:** 64x64 grayscale images of handwritten Kanji characters 
+- **Labels:** 956 unique characters (classes)
+## Model Files
+- **PyTorch Model:** Kanji_ETL8G.pth
+- **ONNX Model:** Kanji_ETL8G.onnx
+- **CoreML Model:** next effort....
+## Usage
+```python
 import torch
-
-# Define the preprocessing function
-def preprocess_image(image_path):
-    image = Image.open(image_path).convert('L')
-    resized_image = image.resize((64, 64))
-    image_array = np.array(resized_image) / 255.0
-    reshaped_image = image_array.reshape(1, -1)
-    return reshaped_image
-
-# Function to predict a label for an image using your PyTorch model
-def predict_label(image_path, model, device):
-    # Convert the preprocessed image to torch tensor and send to the device
-    processed_image = torch.tensor(preprocess_image(image_path), dtype=torch.float32).to(device)
-    
-    # Predict the label using the model
-    with torch.no_grad():
-        outputs = model(processed_image)
-        _, predicted_class = torch.max(outputs.data, 1)
-    
-    return predicted_class.item()
-
-# Create the reverse dictionary for decoding
-index_to_label = {index: label for label, index in label_to_index.items()}
-
-# Test using a sample image
-sample_image_path = ["D:\ETL8G\\0x4e2d\\012780.png", "D:\ETL8G\\0x5e0c\\078406.png", "D:\ETL8G\\0x6d41\\130205.png", "D:\ETL8G\\0x7a40\\046207.png", "D:\ETL8G\\0x7b56\\024793.png"]
-
-for sample in sample_image_path: 
-    predicted_encoded_label = predict_label(sample, model, device)
-
-    # Decode the predicted label using the reversed dictionary
-    decoded_label = index_to_label[predicted_encoded_label]
-    print(f"The model predicts the image label as: {decoded_label}")
-
-
+model = torch.load('Kanji_ETL8G.pth')
+model.eval()
+# Assuming input image tensor is `input_tensor`
+output = model(input_tensor)
+predicted_label = torch.argmax(output).item()
 ```