![]() Scaling the STL model before import in Blender This keeps the units of our model scaled to the Blender reference system, so in later stages the camera has a proportioned size compared to the imported model We click in the File window, then Import, and finally we select the Stl(.stl) option, in order to import the models previously exported from Catia V5.īefore finishing the STL import, we have to specify that the scale would be 0.001, since Blender works in meters and Catia works in milimiters. We then move on to opening Blender to start importing each. It is also possible to create the model from scratch in Blender, however, as we were more fluent in Catia, we decided to do this on that software.Ĭatia environment with the assembly of the objects we aim to recognize In our case, we decided to create the CAD model in Catia V5, and then import it as an. The first step consists of importing (or creating) the objects we want to recognize into Blender. In order to explain how to do this, in this section we’ll walk the reader through the main steps that are necessary to setup a scene that’s compatible with the scripting that will automate the data generation. Therefore, we have to create and setup a scene that tries to resemble the most to the actual, real-life scene in which we would normally find the objects we want to recognize. Whichever it is the object you want to recognize, in order to generate synthĮtic data to train its recognizor, we have to represent this or these objects in Blender. The Blender file explained in this blogpost, as well as the entire code and all necessary ressources can be found here. This software is going to allows to create realistic renderings of the objects seen above, while allowing us to access the position of each object too, a key feature for the labelling to be done. Blender is an open source software used for multiple rendering applications ranging from animation to product design. The algorithm mentioned above is going to be implemented in Python in the rendering software Blender. ![]() ![]() The classes we wish to recognize are the following: In order to do this, we are going to create an algorithm that takes pictures of all of the objects in the same configuration as in the pictures, and also outputs the labels corresponding to the bounding boxes of the location of each object in each image.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |