Image sample obtaining and model training method and system

Abstract

The embodiment of the invention provides an image sample obtaining and model training method and system. The image sample acquisition method is applied to processing equipment in communication connection with a plurality of mechanical arms. the processing equipment comprises a plurality of first special CPUs (Central Processing Units) for reading image data and a plurality of second special CPUs for processing the image data, each mechanical arm is provided with a camera, and the method comprises the following steps: acquiring and storing the image data acquired by each camera in real time; Wherein the plurality of first special CPUs circularly execute reading processing respectively; And allowing the plurality of second special CPUs to perform data processing on the image data read by theplurality of first special CPUs according to a preset data processing mode to obtain an image sample corresponding to each piece of image data. Therefore, the processing equipment adopts a streamingprocessing mode, image data can be processed in real time, image samples can be obtained, a machine learning model does not need to be retrained when new image data are obtained, and the model training time is greatly shortened.

Description

technical field [0001] The invention relates to the technical field of model training, in particular to an image sample acquisition and model training method and system. Background technique [0002] With the continuous promotion of data-driven robotic arm grasping technology, the requirements for the grasping accuracy of the robotic arm are also getting higher and higher. The acquisition of image samples is an important part of data-driven robotic arm grasping technology. Only by obtaining a large number of image samples can an accurate machine learning model for robotic arm grasping be obtained through training of these image samples. [0003] At present, the acquisition methods of image samples generally include: collecting image data through the camera installed on the robotic arm, and processing the image data according to the needs of machine learning model training, such as image color change, matrix transformation, etc., to obtain image samples. After obtaining a la...

AI Technical Summary

Problems solved by technology

[0004] Using the above image sample acquisition method, after the training of the machine learning model is completed, if the camera installed on the robotic arm collects new image data, the new image data must be processed to obtain corresponding image samples and merged into the above-mentioned large number of image samples , and then re-train the machine learning model, which consumes a lot of time for model training

Images