Commit
•
3b0d3ae
1
Parent(s):
ff5990e
Create README.md (#3)
Browse files- Create README.md (2996fa5caac23ec91029c7d4280e53410ca4233e)
Co-authored-by: Renal Shagin <Renaxit@users.noreply.huggingface.co>
README.md
ADDED
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# YOLOv5 Handwritten Text Detection
|
2 |
+
|
3 |
+
Welcome to the Hugging Face repository for the YOLOv5 model specifically fine-tuned for handwritten text detection! This repository, hosted by armvectores, features a state-of-the-art object detection architecture that has been meticulously adapted to recognize and localize handwritten text in images and documents.
|
4 |
+
|
5 |
+
## Model Description
|
6 |
+
|
7 |
+
YOLOv5 is the fifth version of the You Only Look Once (YOLO) object detection algorithm. It excels in speed and accuracy, making it an ideal choice for real-time applications. The YOLOv5 model provided here has been fine-tuned on a diverse dataset of handwritten texts to improve its specificity in detecting handwritten content as opposed to typed or printed materials.
|
8 |
+
|
9 |
+
## Features
|
10 |
+
|
11 |
+
- High Accuracy: Achieves impressive accuracy for detecting various styles of handwriting across different backgrounds and conditions.
|
12 |
+
- Fast Inference: Suitable for real-time applications due to its quick processing time.
|
13 |
+
- Easy Integration: Provides an accessible API for straightforward integration with Python applications.
|
14 |
+
|
15 |
+
## Usage
|
16 |
+
|
17 |
+
To utilize this model for detecting handwritten text in your images, follow the instructions below:
|
18 |
+
|
19 |
+
### Environment Setup
|
20 |
+
|
21 |
+
Ensure you have Python 3.6 or later installed. Then install the required packages:
|
22 |
+
|
23 |
+
pip install transformers torch torchvision
|
24 |
+
|
25 |
+
|
26 |
+
### Inference
|
27 |
+
|
28 |
+
You can perform inference with the following code snippet:
|
29 |
+
|
30 |
+
from transformers import Yolov5Model, Yolov5FeatureExtractor
|
31 |
+
|
32 |
+
# Load model and feature extractor
|
33 |
+
model_id = "armvectores/yolov5_handwritten_text_detection"
|
34 |
+
feature_extractor = Yolov5FeatureExtractor.from_pretrained(model_id)
|
35 |
+
model = Yolov5Model.from_pretrained(model_id)
|
36 |
+
|
37 |
+
# Prepare your image (replace 'path/to/your/image.jpg' with your actual image path)
|
38 |
+
image = Image.open('path/to/your/image.jpg')
|
39 |
+
|
40 |
+
# Perform inference
|
41 |
+
inputs = feature_extractor(images=image, return_tensors="pt")
|
42 |
+
outputs = model(**inputs)
|
43 |
+
|
44 |
+
# Analyze the outputs
|
45 |
+
detections = outputs[0]
|
46 |
+
|
47 |
+
# detections consist of [x_min, y_min, x_max, y_max, confidence, class]
|
48 |
+
# Loop through detections and print results
|
49 |
+
for detection in detections:
|
50 |
+
print(f"Coordinates: {detection[:4]}, Confidence: {detection[4]}")
|
51 |
+
|
52 |
+
|
53 |
+
## Limitations
|
54 |
+
|
55 |
+
While this model performs well across a wide range of handwriting styles, the accuracy may diminish in cases of extremely cursive or overlapping text. The performance is also dependent on the quality of the input images.
|
56 |
+
|
57 |
+
## Contact Information
|
58 |
+
|
59 |
+
For queries regarding this model, please post issues directly on this Hugging Face repository or contact armvectores through their Hugging Face profile.
|
60 |
+
|
61 |
+
---
|
62 |
+
|
63 |
+
This description should provide users with an overview of the YOLOv5 model tailored for handwritten text detection, along with basic usage instructions. Remember, always respect the usage guidelines and terms of service when utilizing models from repositories.
|