SIR: Similar Image Retrieval for Product Search in E-Commerce

Abstract

We present a similar image retrieval (SIR) platform that is used to quickly discover visually similar products in a catalog of millions. Given the size, diversity, and dynamism of our catalog, product search poses many challenges. It can be addressed by building supervised models to tagging product images with labels representing themes and later retrieving them by labels. This approach suffices for common and perennial themes like "white shirt" or "lifestyle image of TV". It does not work for new themes such as "e-cigarettes", hard-to-define ones such as "image with a promotional badge", or the ones with short relevance span such as "Halloween costumes". SIR is ideal for such cases because it allows us to search by an example, not a pre-defined theme. We describe the steps -embedding computation, encoding, and indexing -that power the approximate nearest neighbor search back-end. We also highlight two applications of SIR. The first one is related to the detection of products with various types of potentially objectionable themes. This application is run with a sense of urgency, hence the typical time frame to train and bootstrap a model is not permitted. Also, these themes are often short-lived based on current trends, hence spending resources to build a lasting model is not justified. The second application is a variant item detection system where SIR helps discover visual variants that are hard to find through text search. We analyze the performance of SIR in the context of these applications.

Identical Image Retrieval using Deep Learning

“…The central idea is to convert each image into a fingerprint, signature, or unique descriptor [5]. Internally, fingerprints are basically embeddings calculated from a suitable deep neural network.…”

“…The currently deployed system listens for Kafka themes that stream new and updated images. The running index (last 3 months) of the newly created image is maintained for later search and retrieval [5]. Due to the continuous nature of the application, hash-based indexing is the preferred choice over the technique of collectively learning representations from static datasets such as Principal component analysis (PCA) [3].…”

“…This technology has enabled tremendous growth in markets such as mobile commerce, social networking, and large-scale community projects. In this project, we envisage that such image-based platforms, which focus on image-based search [5,7], discovery, and mobile e-commerce, will present a massive opportunity for new business models.…”