Iterate a Model
When a model is put into use for some time, it might not cover certain scenarios. At this point, the model should be iterated. Usually, using more data to re-train the model can do the job, but it could reduce the overall recognition accuracy and might take a long time. Hence, Model Finetuning can be used to iterate the model while maintaining its accuracy and saving time.
| This feature only works under the Developer Mode. |
General Model Iteration
You can use the following method to fine-tune the model:
-
Acquire images that report poor recognition results.
-
Use Mech-DLK to open the project the model belongs to.
-
Enable the Developer Mode by clicking .
-
Add the acquired images into the training and validation sets.
-
Label the newly added images.
-
In the Training tab, click and then enable Finetune.
-
In the Training parameters tab, lower the Learning rate properly. The Epochs can be reduced to 50–80.
-
Click OK to save the parameter settings.
-
Complete model training and export the model.
Deep Learning Model Iteration
The deep learning models are universal models developed by Mech-Mind. For more information about model introduction and usage instructions, see Deep Learning Model Introduction.
You can use the following method to fine-tune the deep learning models:
-
Acquire images that report poor recognition results.
-
Open Mech-DLK, create a new project, and add the Instance Segmentation module.
-
Enable the Developer Mode by clicking .
-
Add the acquired images into the training and validation sets.
-
Label the newly added images.
-
In the Training tab, click and then enable Finetune.
-
Select Deep learning model finetuning and click
to select the deep learning model (.dlkmp). -
In the Training parameters tab, lower the Learning rate properly. The Epochs can be reduced to 50–80.
-
Complete model training and export the model.
Cascaded Model Iteration
When you train cascaded models, you can select a single module in the Modules area and perform the above operations to iterate the module.
If the previous module is updated, the following module needs to be re-trained. If a module fails to detect or incorrectly detects target objects, the module should be iterated. Then, in the following module, use the newly added images from the previous module as iteration data to iterate the following module.