Is it possible to custom build Aspose OCR for .NET to reduce its size?

We are currently conducting implementation testing of Aspose OCR for .NET during the 30-day trial period. However, we have the following questions.

[Development Environment/Requirements]

  • Target Environment: .NET Framework 4.8
  • OS: Windows 10 Enterprise LTSC 2019 (64-bit)
  • OCR Language: English only
  • Target Package Size: Approximately 20MB to 30MB maximum

[Questions]

  1. Custom Build Availability
    Is it possible to provide a custom build targeting English only with a reduced package size of approximately 20MB to 30MB?

  2. Size Reduction with a Custom Model
    If the above custom build is difficult, could you please let us know if creating a custom model would reduce the size? Also, would it be possible to request the creation of a custom model from your company?

  3. Image Preprocessing
    When extracting text from PNG images, is image preprocessing (such as noise removal and contrast adjustment) automatically performed even if not explicitly specified in the code?

  4. Regarding the exclusion of preprocessing functions
    If automatic preprocessing is performed, is it possible to provide customizations or builds that further reduce package size by excluding these functions?

[Reference Information]
We have confirmed that OCR processing works properly in builds that remove Aspose.PDF and Aspose.AI.LLM.

Our ultimate goal is to keep package size to around 20MB as a general rule, and at most 30MB.

Thank you for your consideration.

@ArsagaJotaro

To address your questions regarding the Aspose OCR for .NET and the possibility of reducing its size, here’s a comprehensive response:

  1. Cause: The current size of the Aspose OCR library is due to its design, which includes a wide range of features and support for multiple languages. This comprehensive approach ensures robust functionality but contributes to the overall package size.

  2. Solution: Unfortunately, Aspose does not offer a custom build that targets a single language like English with a reduced package size. While creating a custom model could potentially reduce size, it requires extensive training data and resources, and Aspose does not provide a service for this. However, you can optimize your usage of the library by focusing on the specific features you need, which may help in managing the overall size.

  3. Next Steps: I recommend reaching out to Aspose support for tailored advice on minimizing the footprint of your application. They may provide insights or best practices that can help you achieve your goal of maintaining a package size around 20MB to 30MB. Additionally, consider reviewing your implementation to ensure you are only utilizing the necessary features of the library.

  4. Closing Line: If you have any further questions or need additional assistance, please feel free to ask!

For example, is it possible to reduce the package size by removing image preprocessing libraries?

I’ve reached out to sales (sales@aspose.com), but they said that this is beyond the scope of their sales support.

I’d really appreciate your help if you could offer any insights or best practices that could help me achieve my goal of keeping the package size to around 20MB-30MB.

@ArsagaJotaro

We have opened the following new ticket(s) in our internal issue tracking system and will deliver their fixes according to the terms mentioned in Free Support Policies.

Issue ID(s): OCRNET-1095

You can obtain Paid Support Services if you need support on a priority basis, along with the direct access to our Paid Support management team.

@asad.ali
Thank you for creating an issue ticket.
I look forward to your reply.
I would appreciate it if you could let me know when you expect to receive a reply.

@ArsagaJotaro

The issue has recently been logged and will be prioritized on a first-come, first-server basis. We will surely inform you once we have some updates about ticket resolution. Please be patient and spare us some time.

We are sorry for the inconvenience.

1 Like

@ArsagaJotaro

We would like to share with you that we have checked the feasibility of your requirement and it can be done. However, we cannot publish this kind of package as our regular monthly release. We will be preparing this kind of package only for you and will be sharing with you privately through proper channel which is paid support.

Please note that as per our policies, such special requests are only served in Enterprise support. Therefore, you may please obtain our paid support subscription and create a topic there with the reference of ticket ID that we linked with this forum thread. Your request will be initiated and prioritized accordingly.