Document import
Arkindex allows you to easily create elements from your images in a pre-defined data structure. This way, you can import a few amount of images either from your computer or from an IIIF server.
Import and organize images of document from files (jpeg, tiff, png), PDF, IIIF manifests.
Annotation
Arkindex allows you to produce the annotations needed to process your documents. Before starting the project, you need to manually annotate a few examples of the expected processing in order to better target the automatic analysis.
Annotate your images with :
- zones of elements on the image, with type and position
- text transcriptions at any level (page, paragraph, line, word)
- classifications
- meta-data
TEKLIA can also provide you with the Callico interface upon request, which integrates with Arkindex for advanced collaborative annotation campaigns.
Processing
Arkindex is a platform for executing any document processing algorithm: OCR, HTR, feature extraction, captioning, translation, etc. Its architecture has been designed to be generic, enabling it to store any type of result, with generic and configurable types.
The following processing types are possible with Arkindex:
Processing type |
Description |
Image Classification |
Associate a class with an image or a portion of an image. |
Object Detection |
Detect an object in an image using a bounding box and identify its type. |
Object Segmentation |
Detect the precise outline of an object in an image and identify its type. |
Image Captionning |
Generate a caption or a tag for an image |
| Transcription | Transcribe printed or handwritten text from an image. |
Classification |
Aociate a class with a text. |
Key-value extraction |
Extract information from an image or text in the form of a key-value association. |
Table Recognition |
Detect and transcribe information presented in the form of a table while preserving its structure. |
Named entity Recognition |
Named entity recognition. |
Entity Linking |
Link a named entity to an existing reference system. |
Translation |
Translate a text from a source language to a target language. |
Geolocation |
Associate GPS coordinates with an image or text. |
Grouping Objects |
Group elements in the same structure. |
Workflow
Arkindex offers extensive capabilities, unmatched by its competitors, for managing complex workflows tailored to your document processing needs:
- Customisable Workflow Design : Arkindex gives you the freedom to define complex workflows tailored to your unique processing requirements. From layout analysis and classification to text recognition (OCR/HTR), named entity recognition and metadata generation, you can curate each step to achieve your desired outcome.
- Real-time monitoring: With Arkindex, you can monitor the progress of each task within your workflow in real time. This powerful feature provides you with an estimated time of arrival for each step, ensuring you can make informed decisions and adjust resources as necessary.
- Error Analysis & Rerun: Not all processes run perfectly every time. Arkindex understands this and provides tools to analyse any errors that may occur in your workflow. Once identified, you can easily rerun processes for those specific elements, ensuring consistency and accuracy.
- Flexible Processing Nodes: To accommodate different infrastructure requirements, Arkindex provides the flexibility to distribute your processing tasks across multiple nodes. Whether it's on-premises, in a cloud environment or even on high performance clusters using SLURM, we've got you covered..
- Integration with custom & open source components: Arkindex is not limited to its built-in functionality. You can effortlessly define your processing steps using your proprietary code or benefit from the vast ocean of open source components available. Docker integration makes integrating these components easy.
