API Docs
Image Text Detection
POST/api/ai/images/v2/image-text-detectionImage Extraction AI
Identifying and extracting text from images using machine learning
ImageCon V2 introduces an advanced text detection and extraction capability that enables the discovery and extraction of machine-readable text from images. This powerful feature opens up new possibilities for image analysis and understanding, with a range of applications across various domains.
One key application is Visual Search, where the detected text in images can be utilized to retrieve visually similar images that contain the same or similar textual content. This enhances search experiences, recommendation systems, and content discovery, enabling users to find relevant visual content based on text-based queries.
In Content Insights, the ability to recognize and extract text from video frames provides valuable information about the themes and topics discussed within the visual content. By analyzing the recognized text, relevant content such as news updates, sports scores, or captions/subtitles can be extracted for further analysis or display, enhancing content understanding and accessibility.
Navigation solutions can leverage text detection to assist visually impaired individuals. By utilizing image recognition, mobile applications can identify and audibly read aloud important textual information such as restaurant names, shop names, or street signs, facilitating navigation and promoting independence for users with visual impairments.
In the realm of public safety and transportation support, ImageCon V2 can detect and extract crucial information like car license plate numbers from traffic camera images. This enables applications such as automated toll systems, traffic violation detection, and vehicle tracking, enhancing efficiency and security in transportation management.
Filtering sensitive information is another valuable application of ImageCon V2's text detection. It allows for the identification and removal of personally identifiable information (PII) from images, ensuring privacy and compliance with data protection regulations.
The text detection feature of ImageCon V2 provides a comprehensive representation of recognized words and lines, highlighting their relationships and positions within the image. This facilitates further analysis, extraction, or visualization of the text, empowering developers to create innovative and intelligent solutions that leverage machine-readable text detection in images.
How does it work?
The text detection and extraction capability of ImageCon V2 utilizes advanced machine learning algorithms and computer vision techniques to identify and extract machine-readable text from images. Here's how it works:
Image Processing: The image is preprocessed to enhance its features and optimize the text detection process. This may involve techniques such as resizing, normalization, and noise reduction to improve the clarity and quality of the image.
Text Detection: The preprocessed image is analyzed using advanced computer vision algorithms to detect regions within the image that contain text. These algorithms are trained on large datasets of images with annotated text regions to learn patterns and characteristics of text.
Text Recognition: Once text regions are detected, optical character recognition (OCR) algorithms come into play. OCR algorithms analyze the text regions and convert the visual text into machine-readable characters. This involves the recognition and classification of individual characters or groups of characters.
Text Extraction: After the text is recognized, it is extracted from the image and organized into meaningful units such as words, lines, or paragraphs. The extraction process takes into account the spatial relationships between characters and the layout of the text within the image.
Text Representation: The extracted text is represented in a structured format, such as a list of words or lines, along with information about their positions and relationships within the image. This representation enables further analysis, extraction, or visualization of the text.
Post-processing and Refinement: To improve the accuracy and quality of the extracted text, post-processing techniques may be applied. This can involve error correction, text normalization, or language-specific processing to ensure the extracted text is accurate and consistent.
Output and Integration: The extracted text is made available as machine-readable data that can be utilized in various applications and workflows. It can be integrated with other systems, processed for further analysis, or used to enhance the functionality of applications that rely on text understanding.
By combining sophisticated image processing techniques, OCR algorithms, and advanced computer vision models, ImageCon V2's text detection and extraction capability provides a robust and reliable solution for uncovering machine-readable text from images. This enables a wide range of applications, from visual search and content analysis to navigation assistance and sensitive information filtering.
Use Cases
The text detection and extraction capability of ImageCon V2 utilizes advanced machine learning algorithms and computer vision techniques to identify and extract machine-readable text from images. Here's how it works:
Image Processing: The image is preprocessed to enhance its features and optimize the text detection process. This may involve techniques such as resizing, normalization, and noise reduction to improve the clarity and quality of the image.
Text Detection: The preprocessed image is analyzed using advanced computer vision algorithms to detect regions within the image that contain text. These algorithms are trained on large datasets of images with annotated text regions to learn patterns and characteristics of text.
Text Recognition: Once text regions are detected, optical character recognition (OCR) algorithms come into play. OCR algorithms analyze the text regions and convert the visual text into machine-readable characters. This involves the recognition and classification of individual characters or groups of characters.
Text Extraction: After the text is recognized, it is extracted from the image and organized into meaningful units such as words, lines, or paragraphs. The extraction process takes into account the spatial relationships between characters and the layout of the text within the image.
Text Representation: The extracted text is represented in a structured format, such as a list of words or lines, along with information about their positions and relationships within the image. This representation enables further analysis, extraction, or visualization of the text.
Post-processing and Refinement: To improve the accuracy and quality of the extracted text, post-processing techniques may be applied. This can involve error correction, text normalization, or language-specific processing to ensure the extracted text is accurate and consistent.
Output and Integration: The extracted text is made available as machine-readable data that can be utilized in various applications and workflows. It can be integrated with other systems, processed for further analysis, or used to enhance the functionality of applications that rely on text understanding.
By combining sophisticated image processing techniques, OCR algorithms, and advanced computer vision models, ImageCon V2's text detection and extraction capability provides a robust and reliable solution for uncovering machine-readable text from images. This enables a wide range of applications, from visual search and content analysis to navigation assistance and sensitive information filtering.
Request
Request samples
Responses
{
"data": {
"lines": [
{
"detected_text": "veniam in proident laboris id",
"confidence": -30895365.34443526,
"geometry": {
"bounding_box": {
"height": -31206125.179928318,
"left": -96637320.12569612,
"top": 72105085.85295844,
"width": -21593577.009167224
},
"polygon": [
{
"x": 63376361.36601299,
"y": -77803673.34713829
},
{
"x": -74521114.12606373,
"y": -26135173.209431574
}
]
}
},
{
"detected_text": "Excepteur pariatur ullamco Duis consequat",
"confidence": -4261648,
"geometry": {
"bounding_box": {
"height": -28190953.19340624,
"left": 89239545.47935206,
"top": -71673093.96241266,
"width": 32407390.979280755
},
"polygon": [
{
"x": -31253958.583303064,
"y": -84851443.30001076
},
{
"x": -84473196.48598465,
"y": 89028023.30285352
},
{
"x": -27443724.96811551,
"y": -48401381.156853415
}
]
}
},
{
"detected_text": "velit magna nisi proident cupidatat",
"confidence": 8906688.849750921,
"geometry": {
"bounding_box": {
"height": 75834347.91818768,
"left": -11821339.03140232,
"top": -56033619.35002953,
"width": 76017443.26157764
},
"polygon": [
{
"x": -77358190.00776175,
"y": 14847286.609454796
},
{
"x": -67689260.22278014,
"y": 77141078.64575005
},
{
"x": -67744233.95401081,
"y": 57326317.07390183
}
]
}
}
],
"words": [
{
"detected_text": "id",
"confidence": -39853875,
"geometry": {
"bounding_box": {
"height": -45470979.08876139,
"left": -47060244.255420305,
"top": 71232768.62651542,
"width": -8867599.416114032
},
"polygon": [
{
"x": 46436426.86077446,
"y": -55724084.958356544
},
{
"x": -64429934.47571213,
"y": -78355501.72030827
},
{
"x": -85404678.74940693,
"y": 54295701.94300091
}
]
}
},
{
"detected_text": "do",
"confidence": -92029197.6418579,
"geometry": {
"bounding_box": {
"height": 91937297.73557901,
"left": 8812557.612716526,
"top": -48544210.269565366,
"width": -59430361.40269524
},
"polygon": [
{
"x": -9339996.09998338,
"y": 10790946.627500102
},
{
"x": 22327456.373290613,
"y": -79893445.13945813
},
{
"x": 61168081.628799826,
"y": -74684930.63338909
}
]
}
}
]
},
"processingTime": 62100747.350598514,
"processingId": "in quis consectetur aute",
"processingCount": 89795562
}