Commodity Registration Using Hand-Aware Discount Seal Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing commodity registration systems fail to appropriately apply price discounts when the discount seal is obscured by a purchaser's hand, leading to potential misapplication of discounts.
Innovation Solution
A system and method that utilize a camera-equipped cart terminal to detect and identify commodities and their attached price discount seals, even when obscured by a hand, by employing image processing techniques such as deep learning and convolutional neural networks to ensure accurate identification and registration of both the commodity and associated discounts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the purchaser holds the commodity in their hand for identification, then the commodity can be recognized, but the price discount seal may be covered by the hand and not detected
Solution Approach 1:
The system performs preliminary commodity identification when the commodity is first held by the hand, then continues to monitor for the seal in subsequent images. This preliminary action ensures the commodity is identified early while allowing continued detection of the seal as the hand moves away.
Solution Approach 2:
The system dynamically switches between two operational modes: when hand-commodity contact is detected, it prioritizes commodity identification; when contact is no longer detected, it prioritizes seal detection. This dynamic adaptation resolves the contradiction by adjusting the detection focus based on the current state.
2Loss of information
If the system prioritizes detecting the price discount seal, then discount information can be captured, but commodity identification may be compromised when the seal is obscured by the hand
Solution Approach 1:
The system dynamically adjusts its detection priority based on whether hand-commodity contact is detected. When contact is present, commodity identification is prioritized; when contact is absent, seal detection is prioritized. This dynamic switching ensures both functions can operate effectively under different conditions.
Solution Approach 2:
The system performs commodity identification as a preliminary step when the commodity is first captured, then focuses on seal detection in subsequent frames. This sequencing ensures commodity identification occurs when visibility is optimal, while seal detection is performed when the hand has moved away.
3Device complexity
If the system uses a single detection mode for both commodity and seal, then the system complexity is reduced, but the reliability of discount application decreases when the seal is covered
Solution Approach 1:
The system uses a unified detection framework that dynamically adjusts its focus based on hand-commodity contact detection. This maintains relatively simple system architecture while improving reliability through adaptive behavior that switches between identification and seal detection modes.
Solution Approach 2:
The single detection system performs multiple functions by adjusting its operational mode: it can identify commodities when held in hand and detect seals when the hand moves away. This multi-functionality within a unified system resolves the contradiction between simplicity and reliability.
Data Source
AI summary
According to embodiments, a commodity registration system comprises a processor configured to: acquire an image from a camera; attempt to detect in the acquired image, a commodity region showing a commodity, a hand region showing a purchaser's hand, and a seal region showing a discount seal attached to the commodity; when a detected commodity region is in contact with a detected hand region in the image, attempt to identify the commodity from the detected commodity region; when the detected commodity region is in contact with the detected hand region in the image and the seal region is detected from the image, attempt to identify the discount seal from the detected seal region; and when the detected commodity region is not in contact with the detected hand region and the seal region is detected from the image, attempt to identify the discount seal from the detected seal region.