Video Screenshot Object Recognition for Faster Commodity Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for finding objects in videos, such as commodities, require complex keyword-based searches that do not allow users to quickly identify desired items, affecting user experience.
Innovation Solution
A method and apparatus that allows users to take a screenshot of an object in a video stream, recognize its type, and trigger an object recommendation control to display a recommendation result page with attribute information of similar objects, simplifying the search process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If keyword-based search method is used to find commodities in videos, then users can search for commodity information, but the operation steps are complex and users cannot quickly find required objects
Solution Approach 1:
The system performs preliminary object recognition and classification on video content in advance, tagging objects with their types and attributes before user interaction. When a user takes a screenshot, the recognized objects and their metadata are already prepared, enabling instant retrieval and recommendation without requiring users to perform keyword searches or navigate through complex steps.
Solution Approach 2:
The patent introduces an intermediary object recognition system that acts as a bridge between the video content and the user's search intent. Instead of users directly searching through keywords, the system automatically identifies objects in screenshots, classifies them, and retrieves relevant information, thereby simplifying the interaction process and reducing operational complexity.
2Measurement precision
If conventional keyword searching and sorted presentation is used, then search results can be presented to users, but users cannot quickly find commodity information meeting their requirements
Solution Approach 1:
The patent replaces the manual keyword-based mechanical search system with an automated image recognition and object classification system. By using computer vision technology to identify and classify objects in video screenshots, the system achieves more precise matching between user intent and commodity information, eliminating the imprecision of keyword searches while reducing the time required to find relevant objects.
Solution Approach 2:
The system changes the search parameter from text-based keywords to image-based object recognition. By transforming the search input from verbal descriptions to visual data, the system achieves more accurate object identification and classification, thereby improving search precision and enabling faster retrieval of relevant commodity information.
3Ease of operation
If users need to open e-commerce application and enter keywords to search, then commodity information can be found, but the operation steps are complex
Solution Approach 1:
The patent merges the video playback function with the object recognition and search function into a single integrated system. By combining these previously separate operations (video viewing, screenshot capture, object recognition, and information retrieval) into one unified process, the system eliminates the need for users to switch between applications and perform multiple discrete actions, thereby simplifying operations and reducing time loss.
Solution Approach 2:
The system implements self-service by automatically performing object recognition, classification, and information retrieval without requiring user intervention. When a user captures a screenshot, the system autonomously identifies the objects, determines their types, and retrieves relevant commodity information, eliminating the need for users to manually enter keywords or navigate through complex search procedures.
Data Source
AI summary
Embodiments of the present disclosure provide a recommendation method and apparatus, a device, a storage medium, and a computer program product. The method includes: obtaining a video screenshot image in response to a screenshot operation on a video stream page; presenting, if the obtained video screenshot image meets a preset condition, the video screenshot image and an object recommendation control on the video stream page, where the preset condition is that an object type of a target object in the video screenshot image is a preset type; and presenting, in response to a first trigger operation for the object recommendation control, a recommendation result page on the video stream page, where the recommendation result page includes attribute information of a recommended object corresponding to the target object.