Systems and methods for utilizing a convolutional neural network architecture for visual product recommendations

A method implemented via execution of computing instructions configured to run at one or more processors and configured to be stored at non-transitory computer-readable media, the method comprising:

• accessing a server storing a plurality of digital images for display on a web site;
• training a two-branch Siamese convolutional neural network (CNN) model by:
• inputting one or more pairs of digital images of the plurality of digital images into the two-branch Siamese CNN model, each respective pair of digital images of the one or more pairs of digital images being pre-labeled;
• and
• determining a respective contrastive loss for each respective pair of digital images of the one or more pairs of digital images, the respective contrastive loss indicating a respective similarity between the respective pair of digital images of the one or more pairs of digital images;
• receiving one or more new digital images;
• and
• determining, using the two-branch Siamese CNN model, as trained, and the one or more new digital images, at least one image of the plurality of digital images to which the one or more new digital images is similar.

A system comprising:

• one or more processors;
• and
• one or more non-transitory storage devices storing computing instructions configured to run on the one or more processors and perform:
• accessing a server storing a plurality of digital images for display on a web site;
• training a convolutional neural network (CNN) by:
• inputting one or more pairs of digital images of the plurality of digital images into the CNN;
• and
• determining a respective similarity for each respective pair of digital images of the one or more pairs of digital images;
• receiving one or more new digital images;
• determining, using the CNN, as trained, and the one or more new digital images, a metric indicating a similarity between at least one image of the plurality of digital images and the one or more new digital images;
• and
• combining the metric indicating the similarity between the at least one image of the plurality of digital images and the one or more new digital images with one or more content-based features to determine one or more second new digital images to which the at least one image of the plurality of digital images is similar.

A method implemented via execution of computing instructions configured to run at one or more processors and configured to be stored at non-transitory computer-readable media, the method comprising:

• accessing a server storing a plurality of digital images for display on a web site;
• training a convolutional neural network (CNN) by:
• inputting one or more respective pairs of digital images of the plurality of digital images into the CNN;
• and
• determining a respective similarity for each respective pair of digital images of the one or more pairs of digital images;
• receiving one or more new digital images;
• determining, using the CNN, as trained, and the one or more new digital images, a metric indicating a similarity between at least one image of the plurality of digital images and the one or more new digital images;
• and
• combining the metric indicating the similarity between the at least one image of the plurality of digital images and the one or more new digital images with one or more content-based features to determine one or more second new digital images to which the at least one image of the plurality of digital images is similar.