Mech Vision

Mech-DLK is part of the machine vision software Mech-Vision and offers a particularly easy-to-learn use of deep learning models.

The deep learning software integrates the entire work process for training deep learning models and makes it possible to train individual neural networks independently, intuitively and quickly or to use already trained models.

This saves time and labour costs, simplifies the process and at the same time significantly improves training efficiency.

In addition, Mech-DLK supports a variety of training models, including defect detection, classification and instance segmentation....

The entire process can be carried out locally by users without external data transfer, providing a high level of data security.

Easy to use

• Even without a technical background, users can train Deep Learning models independently after a simple training session. Without any programming knowledge.

Powerful AI algorithms

• With a variety of powerful AI algorithms, including instance segmentation, image classification, object recognition and semantic segmentation, users can quickly train a wide range of models. This makes it easy to solve the most challenging applications such as overlapping object detection and classification, high-precision measurements and much more.

Efficient training

• With Mech-DLK, users can easily train deep learning models for all types of parts. The high-precision deep-learning algorithms ensure excellent accuracy with fewer required parameters. Advanced data expansion allows users to train a model with smaller image sets. And with the built-in Finetune feature, which dramatically increases training efficiency by optimising existing models, users do not have to train a model from scratch.

Visualised validation results

• Mech-DLK can perform model validation, display validation results, and show the comparison between validation results and annotations. This greatly improves the validation efficiency of vision solutions.

Easy integration

• Mech-DLK supports fast integration