Seamless integration into the MVTec product portfolio
The Deep Learning Tool seamlessly integrates into the MVTec product portfolio with HALCON and MERLIC and serves as the core of your Deep Learning application.
Acquire your images and preprocess them with HALCON or MERLIC if necessary. After labeling, training as well as evaluation in the Deep Learning Tool, deploy your trained network in the respective runtime environment
Discover our Deep Learning Tool leaflet
It provides a general overview on the Deep Learning Tool and its seamless integration with our product portfolio.
The Deep Learning Tool offers
- A fast path to the complete deep learning solution
- An intuitive user interface
- Active support for the optimization of the trained networks
- Easy integration into the MVTec portfolio
- Full control over your own data
Labeling
Data labeling is an essential task for many Deep Learning projects. During labeling, the user adds the information to the system about how the problem is solved correctly. Depending on the method, this information can be image classes, object locations or pixel masks assigned to classes or instances.
Labeling for classification
Labeling for classification is done by simply importing the images and assigning them to a class. If the images are stored in appropriately named folders, they can also be labeled automatically during import.
Labeling for object Detection
With object detection, labeling is done by drawing rectangles around each relevant object and assigning these rectangles to the corresponding classes. Depending on the project requirements, the user can label his data with either axis-parallel or oriented rectangles.
Labeling for segmentation
Labeling for semantic segmentation and instance segmentation can be done by drawing polygonal regions around relevant objects. Labeling for semantic segmentation and instance segmentation can also be done by painting pixel masks with brush and eraser that cover relevant objects. In addition, several smart labeling tools make the labeling process even faster. These tools provide users with instant labeling suggestions – either after selecting a relevant image area or when hovering over an image area.
Labeling for Deep OCR training
By retraining a Deep OCR model, the recognition rate of HALCON’s Deep OCR can be increased even further. With the Deep Learning Tool, large data sets can be labeled very efficiently for this purpose – thanks to the automatic text suggestions of labeled words.
Labeling for Global Context Anomaly Detection
Labeling for Global Context Anomaly Detection is done by simply importing the images and assigning them to respective “good” or “anomaly” classes. If the images are stored in appropriately named folders, they can also be labeled automatically during import.
Training
During training, a pretrained classifier is trained on the image dataset that has previously been labeled. With every iteration over the training dataset, the model tries to improve its predictions measured against the validation dataset. Based on its performance, the weights comprising the neural network are adjusted, improving the performance of the next iteration.
In the Deep Learning Tool, users can set all important parameters in the training page. After selecting a data split, the training can be started and the progress and performance are visualized.
Currently, training can be performed for the following deep learning methods:
- Classification
- Global Context Anomaly Detection
- Object Detection
- Instance Segmentation
- Semantic Segmentation
Evaluation
During evaluation, the model is tested against the test dataset. This step indicates to the machine vision specialist how well the model will perform in practice.
Users can evaluate and compare their trained networks directly in the tool. The evaluation section provides information on model accuracy, including a heatmap for the predicted classes of all processed images, as well as an interactive confusion matrix to help detect misclassifications. Users can also calculate the estimated inference time per image and export the evaluation results as a single HTML page for documentation purposes.
Currently, evaluation can be performed for the following deep learning methods:
- Classification
- Global Context Anomaly Detection
- Object Detection
- Instance Segmentation
- Semantic Segmentation
Version 24.05.1
New Features
- To improve training in very rare cases for Anomaly Detection projects, the option to set the advanced training parameter ‘Weight Prior’ has been added.
- It could happen that the DLT could not be started if another application was installed that used an incompatible version of OpenVINO. This problem has been fixed. Now, the OpenVINO version used by the DLT no longer interferes with other OpenVINO installations.
Resolved Issues and Improvements
- In the Chinese translation of the documentation, an entry in the “Minimum System Requirements” table was missing. This problem has been fixed.
- The status of the Heatmap button on the Evaluation page could be misleading if a project was opened while the list of available DL devices was not ready. This problem has been fixed.
- After optimizing a model for TensorRT or OpenVINO, then switching to Hailo and changing the setting for the option ‘Use allocation script’, the prior optimizations seemed to be lost. This problem has been fixed.
- In GC-AD projects, when drawing a polygon to restrict the domain for postprocessing, the lines were invisible during the drawing. This problem has been fixed.
- When optimizing a model for Hailo, there is now an option that allows enabling a fine-tuning step during optimization. This can improve the performance of the optimized model with the trade-off of a longer optimization time.
- In some cases, the placeholder text in the comment area was not located at the right position. This problem has been fixed.
- If in a GC-AD project from a combined trained model only the local or global subnetwork was selected for optimization, this selection was ignored, and always the combined network was optimized. This problem has been fixed.
- When duplicating a training during a running optimization, the wrong optimization state was shown in the duplicated training. This problem has been fixed.
- If an error occurred during the inference calculation, e.g., due to a missing image, it could happen that the error message was not announced properly. In some rare cases, the DLT could even crash. These problems have been fixed.
- Several clicks on the Duplicate entry in the training item context menu could create several duplicates of the training. This problem has been fixed.
- When selecting images or items on the Evaluation page, unintuitive effects could occur. This problem has been fixed.
- In Semantic Segmentation projects, the optimization of trained models for OpenVINO with precision fp16 failed in most cases. Hence, this option was removed from the precision selection box. Further, in some cases, the precision selection box contained unsupported or not all supported values. These problems have been fixed.
- If the learning rate was changed in the text box and afterward, another training parameter (e.g., number of iterations) was changed using the spin box or mouse wheel, the learning rate changed back to its old value. This problem has been fixed.
- The documentation now provides a more detailed description of anomaly score and anomaly score tolerance.
Contact email for obtaining software installation package: gosoftvip@163.com The resources are sourced from the internet, and this introduction is for learning purposes. If there is any infringement, please contact us for deletion!
For more info write me
Email : gosoftvip@163.com
telegram : @sacyou888