Deep Block is a NO CODE MLOps platform where you can train and use AI models. Deep Block allows you to choose between a library of pre-trained models or to train your own AI, no expertise required.

Deep Block is also the world's fastest imagery analysis solution. Indeed, its patented deep learning technology can quickly process large file sizes (more than 1.2 GB volume imagery within 0.2 seconds).

Deep Block makes for a great solution for remote sensing imagery analysis (aerial surveillance, bi-temporal change detection, object detection and segmentation...) but can accommodate a vast range of use cases like photomicrograph analysis, pedestrian detection, facial expression recognition,...

Coupled with its intuitive click-and-drag interface and growing AI training community, Deep Block aims to become a cross-industrial powerhouse for Computer Vision, Image Recognition & Segmentation and to democratize AI training and use.

Deep Block supports the following image file formats: png, apng, jpg, svg, tiff, bmp, gif, ico and jp2.

Deep Block supports high-resolution files up to 10GB. There is no limit to the number of images you can upload to Deep Block.

The number of annotations needed to train an AI model depends on various factors, such as:

• the complexity of the problem
• the type and quality of data
• the algorithms used 

In general, the more annotations you have, the better the performance of the AI model will be.

For instance, image classification models might require thousands or even millions of labeled images to achieve high accuracy. Object detection models might need hundreds to thousands of labeled images, depending on the complexity and variability of the objects to be detected. Image segmentation models typically require pixel-level annotations for a smaller set of images. 

In any case, it is important to ensure that the annotations are accurate, consistent, and representative of the real-world data to avoid bias and improve the robustness of the AI model. The number of annotations needed to achieve this will depend on the specific use case and the quality standards required.

As a good practice, it is better to test-train your model progressively and validate its performances. If you have any doubt, feel free to contact us directly.

Deep Block enables its users to share projects among themselves.

• Shared projects: Users can share their projects privately with other users directly and invite them to collaborate.
• Project Store: Users can upload their projects to the Project Store for others to use.
• Public projects: Users can make their project public so that other users may offer their help to complete it.

Therefore, the main difference between shared projects and public projects is the willingness to either work in a small private circle or in a more open environment with other Deep Block users.

You can export your projects out of Deep Block directly in a COCO JSON format. You can also use the Deep Block API to integrate directly with your applications. 

You can simply create your account on app.deepblock.net for free. You can also refer to this article: Create an account.

Deep Block offers a monthly subscription ranging from $10 to $300 per month. You can choose a plan depending on your computational needs. 

After cancellation, your projects will be available for the remaining duration of your subscription.

After that, the projects will only be available on the Project Store if uploaded there beforehand. Otherwise, the account will be deactivated and projects deleted.

We recommend you to export your projects before the end of your subscription.