SLAM 3D Bounding Boxes for Deep Neural Network Inventory Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current inventory management methods in retail stores are labor-intensive, error-prone, and delayed, leading to potential lost sales due to misplaced or out-of-stock inventory.
Innovation Solution
The use of SLAM 3D technology to optimize the training and use of deep neural networks for accurate identification and tracking of 3D objects, enabling real-time inventory management and compliance checking on mobile devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual checking of shelves is used, then labor intensity is high and errors are frequent, but the system is simple to implement
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated optical system using mobile device cameras to capture images of shelves. The captured images are then processed by machine learning models to automatically identify and track products, eliminating the need for manual checking while improving accuracy and reducing labor intensity.
Solution Approach 2:
The system enables self-service inventory monitoring where the mobile device autonomously captures images, processes them through deep neural networks, and generates inventory status reports without requiring manual intervention. The machine learning model automatically identifies products and their locations, allowing the system to service itself in terms of data collection and analysis.
2Productivity
If permanent cameras are installed to monitor shelves, then inventory tracking is automated, but the system becomes expensive and complex
Solution Approach 1:
The patent utilizes mobile devices that serve multiple functions: they act as cameras for capturing shelf images, as computing devices for running machine learning models, and as communication devices for transmitting data. This multi-functionality eliminates the need for dedicated permanent camera systems, reducing both cost and complexity while maintaining automation capabilities.
Solution Approach 2:
Instead of installing permanent physical camera infrastructure, the system creates digital copies of shelf scenes using mobile device cameras. These image copies are then processed virtually to extract inventory information, replacing expensive physical infrastructure with software-based processing of digital representations.
3Loss of information
If images are uploaded to central computer for processing, then comprehensive analysis is possible, but significant delay occurs between imaging and corrective action
Solution Approach 1:
The patent performs preliminary processing of inventory analysis directly on the mobile device before returning to the store. The deep neural network models are pre-trained and deployed on the mobile device, enabling immediate processing of captured images during the same shopping trip. This preliminary action eliminates the need to wait for central computer processing, allowing customers to receive instant feedback and take corrective action immediately.
Solution Approach 2:
The mobile device acts as an intermediary between image capture and central computer processing. It performs initial processing locally to provide immediate results, then optionally uploads data to the central computer for further analysis. This intermediary role enables parallel processing paths that reduce overall delay while maintaining comprehensive analysis capabilities.
4Measurement precision
If multiple images of the same object from different perspectives are captured with SLAM 3D marking, then identification accuracy increases, but the complexity of tracking and aggregating results increases
Solution Approach 1:
The patent employs feedback mechanisms where the SLAM system continuously tracks the mobile device's position and orientation, and this feedback is used to adjust the capture of subsequent images. The system uses the identified objects from previous images as reference points to guide future captures, creating a closed-loop system that improves identification accuracy while managing complexity through intelligent feedback-driven sampling.
Data Source
AI summary
A system for tracking of an inventory of products on one or more shelves includes a mobile device. The mobile device has an image sensor, at least one processor, and a non-transitory computer-readable medium having instructions that, when executed by the processor, causes the processor to: apply a simultaneous localization and mapping in three dimensions program, on images of a shelf input from the image sensor, to thereby generate a plurality of bounding boxes, each bounding box representing a three-dimensional location and boundaries of a product from the inventory; capture a plurality of two-dimensional images of the shelf; assign an identification to each product displayed in the plurality of two-dimensional images using a deep neural network; associate each identified product in a respective two-dimensional image with a corresponding bounding box, and associate each bounding box with a textual identifier signifying the identified product.