Method for tracking target object

A non-transitory computer-readable storage medium, on which a computer program is stored, wherein a method for tracking target object is carried out when the program is executed by a processor;

• wherein the method comprises:
• receiving at least one target image captured by a single photon avalanche diode (SPAD) camera before present moment;
• for each target image, inputting the target image and a preset template image into a pre-trained siamese network to acquire a position of a target object in the target image output by the siamese network, wherein the template image includes the target object;
• and
• determining a position of the target object in an image to be predicted based on the position of the target object in each target image, wherein the image to be predicted is an image captured by the SPAD camera at the present moment;
• wherein the inputting the target image and the preset template image into the pre-trained siamese network to acquire the position of the target object in the target image output by the siamese network comprises:
• dividing the target image based on a preset detection area to obtain a detection image corresponding to the target image;
• and
• inputting the detection image and the template image into the siamese network to acquire the position of the target object in the detection image output by the siamese network;
• wherein the inputting the detection image and the template image into the siamese network to acquire the position of the target object in the detection image output by the siamese network comprises:
• inputting the detection image into a convolutional neural network (CNN) in the siamese network to obtain a first high-level feature and a first low-level feature output by the CNN;
• inputting the template image into the CNN to obtain a second high-level feature and a second low-level feature output by the CNN;
• inputting the first high-level feature and the second high-level feature into a cross-correlation convolutional layer to obtain a high-level feature output by the cross-correlation convolutional layer, and inputting the first low-level feature and the second low-level feature into the cross-correlation convolutional layer to obtain a low-level feature output by the cross-correlation convolutional layer;
• and
• determining the position of the target object in the detection image based on the high-level feature output by the cross-correlation convolutional layer and the low-level feature output by the cross-correlation convolutional layer.

An electronic device, comprising:

• a memory on which a computer program is stored;
• and
• a processor configured to execute the computer program in the memory to carry out a method for tracking target object comprising:
• receiving at least one target image captured by a single photon avalanche diode (SPAD) camera before present moment;
• for each target image, inputting the target image and a preset template image into a pre-trained siamese network to acquire a position of a target object in the target image output by the siamese network, wherein the template image includes the target object;
• and
• determining a position of the target object in an image to be predicted based on the position of the target object in each target image, wherein the image to be predicted is an image captured by the SPAD camera at the present moment;
• wherein the inputting the target image and the preset template image into the pre-trained siamese network to acquire the position of the target object in the target image output by the siamese network comprises:
• dividing the target image based on a preset detection area to obtain a detection image corresponding to the target image;
• and
• inputting the detection image and the template image into the siamese network to acquire the position of the target object in the detection image output by the siamese network;
• wherein the inputting the detection image and the template image into the siamese network to acquire the position of the target object in the detection image output by the siamese network comprises:
• inputting the detection image into a convolutional neural network (CNN) in the siamese network to obtain a first high-level feature and a first low-level feature output by the CNN;
• inputting the template image into the CNN to obtain a second high-level feature and a second low-level feature output by the CNN;
• inputting the first high-level feature and the second high-level feature into a cross-correlation convolutional layer to obtain a high-level feature output by the cross-correlation convolutional layer, and inputting the first low-level feature and the second low-level feature into the cross-correlation convolutional layer to obtain a low-level feature output by the cross-correlation convolutional layer;
• and
• determining the position of the target object in the detection image based on the high-level feature output by the cross-correlation convolutional layer and the low-level feature output by the cross-correlation convolutional layer.