How It Works
ANSTS will automatically validate the correctness of data, including image and label files. It also provides a built-in labeling tool, allowing users to perform labeling tasks and convert labels if necessary.
Product Features
No coding
ANSTS is specifically designed for engineers who do not have a profound knowledge of programming skills to design a proper object detection model in only a few simple clicks. It provides a graphical user interface (GUI) that makes it easy to select the appropriate components, configure the model, and train it.​​​
Built-in labeling tool
The ANS Labeling Tool enables you to easily add, modify, and delete dataset labels in the ANS format. You can import label files from formats like YOLO, Pascal VOC, and JSON, converting them to ANS format while preserving your original label data.​
The tool supports two labeling types:
-
Bounding Box: Draws rectangles around objects, providing an approximation of position and size.
-
Polygon: Ideal for complex shapes, offering precise boundaries for objects that cannot be accurately captured by a rectangle.
Semi-Automatic Labelling
The ANS Labeling Tool provides two semi-automated labeling methods to enhance your workflow:
-
Auto-Labeling: This feature automatically labels all images using AI inference with a pre-trained model, creating object classes and bounding boxes. You can review and modify the labels as necessary.
-
Auto Object Detection: This method automatically highlights objects in images by drawing bounding boxes, enabling you to quickly assign object class names. This approach saves time compared to manual labeling.
Simplify Model Training
ANSTS utilizes NVIDIA GPUs to speed up training and process data locally, ensuring data privacy. Its user-friendly interface allows for easy adjustments of training parameters, making it a convenient option for object detection projects. The ANSTS training algorithm features an early stopping mechanism, meaning it will halt if no further improvements are detected. Additionally, ANSTS provides a training and validation graph, allowing users to make informed decisions about whether to continue or manually stop the training process.
Model evaluation & testing
The built-in evaluation and testing interface allows users to test the performance of the models before final deployment. Users can select any image dataset or built-in video capture devices in their system to test how the trained model performs object detection tasks in real time.
Optimize for Deployment
ANSTS provides developers with the capability to export their trained models in optimized formats (GPU and CPU). These formats are specifically designed for seamless integration, allowing the models to be utilized directly within LabVIEW. Additionally, the exported models can be deployed across various hardware platforms utilizing ANSVIS, ensuring flexibility and compatibility in different environments. This functionality enhances the efficiency and effectiveness of implementing machine learning solutions in diverse applications.
