In traditional Vision application field, OCR task is very popular and most of time consuming is optimize parameters, in which the text could be well segmented from background and recognized accurately. However, sometime its is impossible to read correctly in case of various sized, non-uniform backgrounds, blurred characters, reflective...
Printing quality is the first priority for every printer manufacturer, and one of most important part inside one printer is roller. Unfortunately, roller is made from a material which easily get defect such as hole, burn, scratch...Due to various of form and dark, those defects mostly still are checked...
For Pick and Place application, Adaptive Vision Studio (AVS) easily integrate with almost any robot through some common communication interface such as TCP/IP, in which AVS will return the object location in world coordinate within few ten ms, thus can meet up the high speed requirement
Most of traditional approach for OCR application is using SVM to classify, it works well with printing document due to the font, color and size not change too much. When facing to hand writing document, this approach become less accuracy and almost cannot use in production. Thanks to Deep...
