Shelf Video Training Model for Product Takeout Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting customer actions on products in retail environments, such as taking a product from a shelf, face challenges in accuracy and efficiency, particularly when dealing with complex scenes and multiple objects.
Innovation Solution
A system that generates combined image data by arranging object images near the person in the image, using a machine training model to enhance detection accuracy through contrast training and machine learning techniques, specifically involving a backbone, adder, encoder, and decoder architecture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine training model is used to detect customer actions on products, then detection accuracy is improved, but the complexity of the system increases
Solution Approach 1:
The system segments the detection task into multiple specialized components: a backbone network for feature extraction, an adder module for integrating multiple feature representations, and a detector for action recognition. This segmentation allows each component to be optimized independently, improving overall detection accuracy while managing system complexity through modular design.
Solution Approach 2:
The adder module serves as an intermediary between the backbone feature extraction and the final detection stage. It integrates multiple feature representations (spatial, temporal, and contextual features) before passing them to the detector, thereby improving detection accuracy while maintaining a clear architectural structure that manages complexity.
2Measurement precision
If detailed detection rules are set according to camera arrangement and person direction, then detection accuracy is improved, but the difficulty of detecting and measuring increases
Solution Approach 1:
The system changes parameters dynamically based on camera arrangement and person direction. The backbone network adjusts feature extraction parameters, and the adder module integrates multiple views with different parameters, allowing the system to maintain high detection accuracy across various camera configurations without requiring manual rule adjustments for each scenario.
Solution Approach 2:
The machine training model is designed to be universal and adaptable to different camera arrangements, person directions, and product types. By training on diverse data and using a flexible architecture with the adder module that can integrate multiple feature types, the system achieves high detection accuracy without requiring separate detection rules for each specific scenario.
3Measurement precision
If combined image data is generated by arranging object images near the person, then the machine training model's ability to distinguish actions is improved, but the processing time increases
Solution Approach 1:
The system performs preliminary actions by pre-processing video data to generate combined image data that arranges object images near the person before the main detection process. This pre-arrangement of visual information in the combined image data enables the backbone network and adder module to quickly identify actions without requiring extensive processing during the main detection phase, thereby improving action identification accuracy while managing processing time.
Data Source
AI summary
A non-transitory computer-readable recording medium stores therein a generation program that causes a computer to execute a process including acquiring a video obtained by imaging an area including a product shelf on which products are arranged; specifying an action of a person holding the product by analyzing the acquired video, specifying an image frame including a product stored on the product shelf and a product held by the person from a plurality of image frames that form the acquired video based on the specified action of the person holding the product, and generating a machine training model trained to identify a person performing an action of taking out the product from the product shelf using the specified image frame.