Developers can leverage ProductAI to productively build high-performance, scalable visual recognition enabled applications via open source SDKs or RESTful API. The platform offers best-in-class visual product recognition (e.g. search and tag retail goods, including fashion, textiles, furniture, wine, etc.) and general visual recognition (e.g. search and tag any image). The computer vision tasks include search, detection, classification, tagging, and analysis. ProductAI is a mature computer vision cloud platform built on years of R&D by Malong Technologies; a Softbank-backed company with AI that holds first place wins from Google, Microsoft and Amazon (learn more at This portal,, is the central place for all ProductAI technical documentation.

ProductAI helps developers build an image search engine based on their own images in the cloud. Image search, sometimes referred to as visual search or image retrieval, is the capability to obtain a ranked list of candidate images best matching a query image. What’s special about ProductAI is scenario-based image search. That is, there are multiple pre-defined scenarios for image search which can be selected when using ProductAI. By selecting the appropriate scenario, image search accuracy can be greatly improved because the underlying models are tuned for that vertical, such as apparel, furniture, or wine. The process of uploading, indexing, and querying images are made easily accessible via the ProductAI web console, SDK or API.

Object Detection

Object detection refers to the computer vision task of locating and classifying objects within images. ProductAI offers state-of-the-art object detection services, most of which are product related. For example, using ProductAI, developers can enable their applications to “see” within images to detect products such as electronics, apparel, and furniture. In addition to products, ProductAI offers other types of detection that may be used in conjunction with product recognition applications, such as text (which may be recognized on products) or people (who may be wearing or interacting with products). Developers may leverage ProductAI object detection by uploading a query image through its web console, SDK or API. The response from the service indicates the likely location and category information of the objects detected.

Image Annotation

Automatic image annotation is sometimes referred to, or is related to, tagging, analysis, attributes, or image understanding. This computer vision task interprets the content of an image and outputs semantic information, in the form of one or more labels. The usefulness of this technology can be seen in applications that need to sort, group, describe, or filter images. ProductAI provides world-class image annotation to developers in various levels of semantics: from the general domain, for example, general tags and color attributes, to vertical domains, for example, fashion product analysis. Developers can leverage automatic image annotation through the web console, SDK or API.

Combined Services

A concept from ProductAI to highlight is “Combined Services.” It is an optional capability offered by the platform which, in certain use-cases, significantly improves efficiency. Combined Services enable developers to select multiple ProductAI APIs and compose them into a single API call. That is, instead of making calls one by one, a developer can send a predefined call graph, which ProductAI will execute internally and aggregate sub-results into a unified response. It can save developers from implementing some common API logic at client side, which reduces potentially numerous round-trip calls into just one.