sodai.vision

What It Is

SODAI Vision API is a bulky waste image recognition API operated by Japan’s オークネット（Aucnet）. It is designed for local governments’ bulky waste intake workflows: after a user takes a photo of an item they want to dispose of with a smartphone and sends it in, the AI can determine whether the item can be processed as waste and what handling fee applies. The page states that the City of Yokohama’s bulky waste reception chatbot has already adopted this API, and it also provides a LINE@ integration demo.

Core Capabilities and Integration

Its AI capabilities are based on deep learning. Rather than general-purpose visual recognition, the focus is vertical recognition around “bulky waste item categories.” The provider says it has analyzed bulky waste item categories from municipalities across Japan and built common master data, while also supporting additional learning for region-specific item categories. This is valuable in a market like Japan, where municipal rules can vary significantly. On the development side, the page says developers can obtain a Token through the SODAI management console to call the API, and it provides API specifications as well as a list of items that can be identified. The LINE@ demo suggests it is well suited for embedding into chatbots or mobile self-service consultation flows.

Pricing and Trial

Pricing information is limited. The page clearly offers a “トライアル（接続確認用）” plan, marked as free of charge; after application, users start by default with this connection-verification trial. There are also “トライアル（事前検証用）” and “スタンダード（本番用）” options, but specific pricing, request volumes, and production deployment costs all require inquiry through the help channel.

Pros and Cons

The main advantages are its clearly defined use case, an existing Yokohama City deployment, and a large listed set of recognizable bulky and non-bulky waste items, making it easier for buyers to evaluate initial coverage. The ability to add learning for regional item categories also improves real-world adaptability. The drawbacks are that the page does not provide key metrics such as recognition accuracy, confidence scores, response speed, SLA, image data retention, or privacy policies. Chinese-language support is also not mentioned.

Who It’s For and Access from China

It is best suited to Japanese municipalities, government service contractors, waste intake systems, and chatbot developers. If used for waste sorting or urban government service scenarios in China, local item categories, fee rules, language support, and compliance requirements would need to be reassessed. The page provides no information about access from mainland China, and payment methods are not disclosed. Alternatives include image recognition services from Baidu, Alibaba Cloud, and Tencent Cloud, or general-purpose vision APIs such as Google Cloud Vision, AWS Rekognition, and Azure AI Vision.