Iterate a Model
If the model performs poorly after deployment, it can be improved through iteration. 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
Models you trained can be fine-tuned using the following steps:
-
Acquire images that report poor recognition results.
-
Use Mech-DLK to open the project the model belongs to.
-
Enable the Developer mode by clicking
. -
Import the collected image data and complete labeling.
-
Click
under the Training tab. On the Model finetuning tab of the Training Parameter Settings window, enable Finetune and select Self-finetuning.
-
Navigate to the Training parameters tab, lower the Learning rate, and choose 50 to 80 for the Epochs.
-
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 corresponding module.
-
Enable the Developer mode by clicking
. -
Import the collected image data and complete labeling.
-
Click
under the Training tab. In the Training Parameters window, go to the Model finetuning tab, enable Finetune, check the Deep learning model finetuning option, and then click
to select a deep learning model (a .dlkmp file).
-
On the Training parameters tab, lower the Learning rate properly. The Epochs can be reduced to 50–80.
-
Complete model training and export the model.
Cascade Model Iteration
When you train cascade 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 escapes and overkills exist in a module, 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.