Object location method, device and storage medium based on image segmentation

An object location device based on image segmentation, wherein the device includes a processor, a memory, and a program of the object location based on image segmentation stored on the memory and operable on the processor, the program of the object location based on image segmentation being executed by the processor to implement the following method:

• Collecting training images and labeling the training images to obtain a trained database;
• Designing a fully convolutional neural network;
• training the fully convolutional neural network to obtain a target neural network, by inputting the trained database into the fully convolutional neural network;
• Labeling and locating object images, based on the target neural network;
• the fully convolutional neural network, includes convolution layers, pooling layers and deconvolution layers;
• the training the fully convolutional neural network to obtain a target neural network, by inputting the trained database into the fully convolutional neural network, further comprises:
• obtaining, a target deconvolution layer, by inputting instructions into the fully convolutional neural network to train the deconvolution layer, based on the output of the fully convolutional neural network;
• wherein the target neural network including the target deconvolution layer, and labeling and locating object images, based on the target neural network, further comprises:
• obtaining, a plurality of pixels of the object images, through a down-sampling of the target neural network;
• classifying, each pixel in the target neural network, to obtain classification output of the plurality of pixels;
• obtaining, classified images, by a up-sampling of the target deconvolution layer, the classified images having the same size as the original object images.