CG generation for defect classification (teacher images for machine learning)

basic information

● Any material can be set Materials that can be scratched can be freely configured. For example, by replacing textures such as ceramic, metal hairline finish / satin finish, it can accommodate any material. By creating textures and shapes, it can also handle complex materials like silicon wafers. ● Types and methods of scratches are completely customizable By freely setting parameters, it generates all shapes of scratches in terms of size, length, depth, position, angle, etc. In addition to basic patterns like scratches, dents, and cracks, anything that can be represented in 3D can be customized. It also supports product image representations such as bottles and smartphone cases. ● More realistic with light source settings Since all materials and scratches are represented in 3D shapes, they are reproduced with the correct appearance relative to the light source. By customizing the number, color, irradiation range, angle, etc., of the light source settings, it can closely resemble real environments. ● High-speed generation of tens of thousands of training data With automatic labeling and random output functions, it is possible to generate thousands of images per hour in one step.

Price range

Delivery Time

Applications/Examples of results

Appearance inspection

Detailed information

• 

  Any material can be set. Materials that can cause scratches can be freely configured. For example, by replacing textures such as ceramic, metal hairline finish, or satin finish, it can accommodate any material. By creating textures and shapes, it can also handle complex materials like silicon wafers.

• 

  Types of wounds, methods of infliction, and shapes are completely customizable. By freely setting parameters, you can generate all kinds of wound shapes, including size, length, depth, position, and angle. In addition to basic patterns like scratches, dents, and cracks, anything that can be represented in 3D can be customized. It also accommodates product image representations such as bottles and smartphone cases.

• 

  High-speed generation of tens of thousands of teacher data. With automatic labeling and random output functions, it is possible to generate thousands of images per hour in one step.