Update README.md
Browse files
README.md
CHANGED
|
@@ -53,6 +53,7 @@ This model addresses the challenge of digital recognition for the Geez script by
|
|
| 53 |
|
| 54 |
The model is intended for direct use in digitizing handwritten Geez documents, educational language learning tools, and automated data entry systems. Users input a cropped image of a handwritten character, and the model returns the predicted character class and confidence score.
|
| 55 |
|
|
|
|
| 56 |
### Downstream Use [optional]
|
| 57 |
|
| 58 |
<!-- This section is for model use when fine-tuned for a task, or when plugged into a larger ecosystem/app. -->
|
|
@@ -84,6 +85,20 @@ The model is **not** designed for:
|
|
| 84 |
|
| 85 |
Users should implement a pre-processing pipeline to segment words into individual characters before feeding them into this model. Images should be normalized to 128x128 pixels and converted to grayscale.
|
| 86 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
## How to Get Started with the Model
|
| 88 |
|
| 89 |
Use the code below to get started with the model.
|
|
|
|
| 53 |
|
| 54 |
The model is intended for direct use in digitizing handwritten Geez documents, educational language learning tools, and automated data entry systems. Users input a cropped image of a handwritten character, and the model returns the predicted character class and confidence score.
|
| 55 |
|
| 56 |
+
|
| 57 |
### Downstream Use [optional]
|
| 58 |
|
| 59 |
<!-- This section is for model use when fine-tuned for a task, or when plugged into a larger ecosystem/app. -->
|
|
|
|
| 85 |
|
| 86 |
Users should implement a pre-processing pipeline to segment words into individual characters before feeding them into this model. Images should be normalized to 128x128 pixels and converted to grayscale.
|
| 87 |
|
| 88 |
+
## Evaluation
|
| 89 |
+
|
| 90 |
+
### Testing Data, Factors & Metrics
|
| 91 |
+
|
| 92 |
+
#### Metrics
|
| 93 |
+
- **Accuracy**: The primary metric used for evaluation.
|
| 94 |
+
|
| 95 |
+
#### Inference Performance
|
| 96 |
+
- **Single Image Inference**: 81% baseline accuracy.
|
| 97 |
+
- **Test-Time Augmentation (TTA)**:
|
| 98 |
+
- Configuration: 10 augmentations with majority voting.
|
| 99 |
+
- Result: Achieves approximately **90% classification accuracy**.
|
| 100 |
+
- Impact: Significantly reduces error rates caused by handwriting variability.
|
| 101 |
+
|
| 102 |
## How to Get Started with the Model
|
| 103 |
|
| 104 |
Use the code below to get started with the model.
|