FORXAI Visual Inspection


FORXAI is an image IoT platform that accelerates the DX of society together with customers and partners.

Two thoughts are included: “Foresight: Foresight” and “AI for society: For X AI”. The overlapping arrows with “X” meaning change represent the future-oriented progress of the fusion of customers, partners, and our technology.

FORXAI Visual Inspection (VI)

Zero-Code Platform

One of the obstacles of any organization embarking on AI technologies is usually the lack of resources, especially in the areas of deep learning. Skilled & experienced resources is currently scarce, and the cost can be quite high if we factor the technology associated with it.

However, in order for businesses to stay relevant, we cannot keep putting the adoption of AI to the back burner. A solution is to use tools that offer less reliance on this skillset and anyone in the organization can contribute to the building of an AI model. This is where FORXAI VI (Visual Inspection) comes in.

FORXAI VI (Visual Inspection) is a platform that enables machine vision and machine learning users, whether you are an engineer, QC/QA inspector, data scientist, to easily collaborate & build an AI inspection or monitoring model without a single line of code. Check out how it works below:

How it works

The project in FORXAI VI determines the type of problem we are solving. For example, we may be trying to classify images (e.g. Cats, Dogs images etc), or trying to find what & where is the anomaly, or detecting objects and its position, or even separating an object from its background.

With simplified icons and description, it makes it easy for anyone to decide on which model to create.

Data is the most valuable asset in Machine Learning. Without good data, the success of the project will diminish. Here, we make it easy to import your data (images), just drag & drop, no additional meta-data to input.

This is where you tell the system what is the area of interest & where is it. This is one of the most time-consuming and tedious work but also the most important task. Accurate labelling will yield a good AI model. To help users in this process, the annotation tool provides features to zoom, pan & navigate to each image.
Need to speed up? Use the Auto-Annotation feature. The system will predict where to annotate based on your previous input.

This interface allows users to set training parameters prior to model training, such as splitting of Training/Test ratio, detection threshold, add augmentations to create more scenarios for training, or selecting if you want a small or big model (small models typically for test and big models for production).

Realtime loss curves & model metrics provides users with snapshots of the training process & if further training is required.

Evaluation metrics provides users with the results of the trained model. You can also compare results of previously trained models to select the best one before deployment.

We offer a simple way to deploy your model to a FORXAI VI Runtime by easily entering the IP address of the destination station.
Once deployed, you have successfully served a model.

What’s Next?

Consume your Model

Consume your newly created model through API’s and integrate it into your process. This may involve integration to new or existing cameras, controllers, systems etc. depending on your environment & how automated your processes are.

If this is too complex, contact us for more information. We can provide the end-to-end deployment including integration to various equipment or systems.

Key Benefits

• You do not need to be an AI-expert to build an AI model

• Reduce deployment time

• Reduce reliance on man-power

• Increase product quality & reduce waste

• Upskill workers from manual inspectors to data-driven workers

Applicable Scenarios

Besides using it for Visual Inspection, you can extend FORXAI Visual Inspection to be the eyes in your shop floor and link it to Autonomous Mobile Robots for a total integrated solution.