Text Detection

This function block is used to identify and locate text within images using a deep learning model. It processes the input image and provides outputs indicating where text is detected.

📥 Inputs

Image Any The input image in which text needs to be detected.

📤 Outputs

Image Any The modified image with detected text locations highlighted.

Referance Point The coordinates of the reference point for each detected text box.

Referance Rectangles The coordinates of the rectangles bounding each detected text area.

Number of Detected Text The total number of text areas detected in the input image.

🕹️ Controls

Confidence A slider to set the confidence threshold for text detection. Higher confidence values lead to stricter detection criteria.

NMS Threshold A slider to adjust the Non-Maximum Suppression (NMS) threshold, which helps filter out overlapping bounding boxes.

🎨 Features

Deep Learning Model Utilizes a pre-trained deep learning model for robust text detection.

Real-time Feedback Adjusting controls allows users to modify detection settings in real time, improving detection results based on the input images.

📝 Usage Instructions

1. Input Image: Connect an image source to the Image Any input.

2. Adjust Detection Settings: Use the Confidence and NMS Threshold sliders to configure the detection parameters.

3. Run the Block: Execute the block to perform text detection. The output image will highlight detected text areas along with their reference points and the quantity of detected text.

📊 Evaluation

Upon evaluation, this function block processes the input image to detect text and outputs a modified image with detected text highlighted, as well as additional information such as reference points, rectangles, and the count of detected text areas.

💡 Tips and Tricks

Improving Detection Accuracy

Use higher confidence values (e.g., above 70) for clearer text to ensure only the most confident detections are returned.

Handling Different Lighting Conditions

If text detection is poor in low light or uneven lighting, consider preprocessing the image with an Auto Contrast or Histogram Equalization to improve visibility.

Selecting Input Types

Ensure that the input image is of good quality with clear text for optimal detection results. Low-resolution or blurry images may hinder performance.

🛠️ Troubleshooting

No Text Detected

If no text is detected, ensure the image quality is high, and verify that the input contains legible text. Adjust the Confidence slider to a lower value to test for more sensitive detection.

Performance Issues

For larger images, reducing the input dimensions may improve processing speed. Consider resizing images if you encounter slow performance.