Visual AI solutions by Ximilar are based on the technologies of machine learning and computer vision. Machine learning describes a process in which a computer learns from experience without being explicitly programmed. Computer vision enables computers to extract meaningful information from images, videos, and other visual inputs. Based on these principles, we have built a visual AI platform accessible via web application and API.

The image processing is performed by separate tasks. Your task is in fact the last, and therefore the most accurate, version of the machine learning model you trained for specific purposes. For example, when you want to build an object detection solution for identifying cars in CCTV footage, you train an object detection task to detect cars and a separate task to categorize them, based on their type or brand.

There are various ways to test the accuracy of your tasks, and based on the testing you can modify and re-train them. You can then combine them in a Flow and deploy them as a complete solution available through API.

This way, we are able to fully automate the processing of customers’ visual data, such as recognition, categorization and tagging of images, object detection, image matching, visual and similarity search, background removal, resolution enhancement, and so on.


Go to:

Ximilar’s AI solutions were built to handle large datasets containing millions of images. Our cloud doesn’t physically store image collections for visual search. We process each image once during synchronization, extract descriptive data, and then immediately discard it, allowing for swift image processing. For instance, our largest collections, holding over 100 million images, require only a few hundred milliseconds to search for visually similar items.

Complex images containing more objects, patterns, topics, people or items, can be easily processed with our AI. For instance, our Fashion Search and Home Decor Search solutions automatically detect all fashion apparel or homeware in the images, providing categories and similar stock to each object individually. You can test these solutions in our public demo.

We also enable asynchronous requests. Read more about them in the API Documentation.

 


Go to:Read more:

As of now, the way to delete your account in our App is to write us an e-mail at with a request to delete your account. After confirmation, your account will be deleted together will all your data.

Read more:

This depends on how you use the solution and how it is deployed. If you are using our standard API and your collection is synchronized on the cloud, then you need an internet connection. However, you can also use an offline model, or deploy it on your infrastructure. This is useful for companies without stable internet connections. Don’t hesitate to contact us to discuss your needs.


Go to:Read more:

If your mobile app runs online, it can be connected to our standard API and the model can be deployed on the cloud. We can even prepare your own cloud.

Ximilar offers offline deployment of solutions created on our platform. The model can run on any workstation with Windows, Linux, or MacOS or a modern smartphone with Android or iOS. The model can be downloaded in several formats (TensorFlow, OpenVINO, TFLITE) and used on mobile phones, edge devices, NVIDIA Jetson, or your workstation.

Read how we prepared a custom object detection solution on NVIDIA Jetson Nano. Contact us to discuss your use case and prepare a custom solution.


Go to:Read more:

Ximilar systems can be deployed on your private clouds, or we can create a dedicated instance for you. This is great if you have large volumes of data to be processed or for specific security reasons. It can be done in a region of your choice. Privacy regulations are then applied according to that region’s rules (e.g., the USA vs. European laws). Contact us to discuss your needs and learn more.


Go to:Read more:

The probability of your service being discontinued is very low. However, if this situation happens, you will get all custom models trained on your data (e.g., classification, detection, custom similarity models) in standard ML formats (e.g., TF Lite, TensorFlow) and you can run them on your hardware platform.

 

Read more:

You can connect to the Ximilar API with any programming language or with a curl command. You can also test the API with Postman or Insomnia. Contact us if you need help with the setup.


Go to:

An authentication token is secure identification of your account. It allows you to access Ximilar API services. Each request to the Ximilar API must contain your authentication token in the request header. Find more in our API documentation.


Go to:

Yes, we are actively using and developing SDK library in Python. You can also query any uploaded and annotated data via standard REST API requests. Simply call any endpoint that is described in our API documentation.


Go to:

By using online solutions built with our platform, you make API calls (requests). Different API calls have different values in API credits. These values are listed and explained in our Pricing and API documentation.

Every pricing scheme, including the Free plan, has a monthly supply of API credits. You can keep track of your credit consumption in the App, and you will be notified via e-mail when 70, 90, and 100 % of your credit supply runs out. The unused credits from the monthly supply expire by the end of the monthly period.

Apart from the monthly supply, you can purchase as many API credit packs as you need. These credit packs will only be used after your monthly supply runs out, and will therefore secure the consistency of your service in case of unexpected system loads. Unlike the monthly supply, the API credit packs do not expire.

We can also discuss a custom pricing and API credits scheme tailored to your business’s needs. Feel free to contact us.


Go to:Read more: