Model Iteration
After a certain period of usage, you may find that the trained model may not be applicable to certain scenarios. At this point, you should iterate the model. It is common practice to re-train the model with more data, but such an effort could reduce the overall recognition accuracy and might take a long time. Hence, it is recommended to use the Model Finetuning function for model iteration so as to maintain accuracy and save time.
General Model Iteration
-
Collect 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 collected images into the training and validation sets.
-
Label the newly added images.
-
Click and then enable Finetune.
-
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.
Super Model Iteration
-
Collect 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 collected images into the training and validation sets.
-
Label the newly added images.
-
Click and then enable Finetune.
-
Select Super model finetuning and click
to select the super model. -
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.