Warning
You are reading an old version of this documentation. If you want up-to-date information, please have a look at 5.0 .3.2 Image Database Creation
![../../../../_images/candidate_selection_image_database_creation_finished.png](../../../../_images/candidate_selection_image_database_creation_finished.png)
Fig. 78 Candidate Selection - Image Database Creation finished
In order to feed the EYE+ neural network and train it to learn what differentiates a good from a bad candidate you need to generate an image database with parts with different orientations and positions on the production plate. You must provide at least 50 candidates in the database to proceed to the next step.
Important
Do not use the same image in the database twice. Apply vibration to flip the parts between each image acquisition or move them manually on the platform.
Note
You can add more than 50 candidates to the database if necessary or come back later to this step to add more images.