Frequently Asked Questions

What is Macula AI?

An integrated solution based on advanced artificial intelligence to automatically detect and identify defects on manufacturing lines. It can also monitor, make predictions and generate reports to improve the quality of the products and reduce loss.

What are the advantages of using AI and Deep Learning for visual inspection?

On standard datasets, computer vision has outperformed human accuracy specially with low quality images. There are several approaches to explain the model, so we can understand why it decided to make its prediction. Macula AI goes beyond traditional visual inspection methods, which are non-automated and difficult to adapt to the growth of today's manufacturing production.

I want to use Macula AI to automate my visual inspection process. How do I get started?

There are 3 simple steps to get started with Macula AI:

1. Trial

2. Validation

3. Operation.

First, our team of machine learning experts will provide an AI model tailored to meet your needs and objectives for you to try within a few days. The AI model is then adjusted to your specific use cases for the validation period. Once the AI model is approved, our team will connect Macula AI into your facilities.

Does it require training and/or machine learning expertise to install and operate Macula AI?

Intended for companies that do not benefit from a team of data scientists, Macula AI allows you to easily operate advanced visual recognition systems, from deployment to daily use. These companies can thus ensure their competitiveness on the market and fill the resource gap.

How much data do I need to use Macula AI?

Macula AI uses state-of-the-art learning techniques like self-supervised learning and zero-shot learning to understand the actual notion of the defect. Our training and inference framework allows us to learn collectively from synthetic data along with the actual data. Because of that, our models can generalize with limited amounts of real data. We have a very strong data augmentation and preprocessing framework, which allows us to substitute the required quantity by resampling from available data points.

What is synthetic data?

Synthetic data is an innovative and unique approach designed to fill the data gap often encountered by our clients. This technique generates the adequate data quantity to train the Deep Learning algorithms in order to achieve accurate results.

How is the data cleaning and data annotation handled?

It depends on the complexity of the data. We have certain tools in place which require minor assistance from our team members combined with the active learning technique to partially automate these tasks. If the data is too complicated, the data cleaning and annotation can be done manually by our professionals.

What cameras can I connect to Macula AI?

Macula AI was designed to be hardware agnostic. Hence, Macula AI is compatible with any cameras with a standard connectivity such as: RGB cameras, multispectral cameras and thermal cameras.

Where do I store my images?

Customization and safety being important components in the design of Macula AI, client’s data is stored in the cloud, either directly onsite or on peripherals, depending on the infrastructure of the institution.

Does Dataperformers own my data?

No. When you use the services of Dataperformers and its products, you remain the sole owner of your data. Under no circumstances will they be used for purposes other than those agreed upon in the use of the products.

Get started with Macula AI

Thanks for submitting!