On-Device AI Product Recognition With Server Model Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional online product purchase systems are cumbersome and inconvenient for offline product searches, with delayed response times and reduced recognition rates due to high data transmission and computational demands, especially when recognizing new products and real-time updates.
Innovation Solution
A mobile artificial neural network device equipped with a camera, AI recognition model, and processor that performs real-time product recognition and augmented reality display, minimizing data transmission and computational load by processing AI recognition locally and updating models through a server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI recognition model operations are performed on a server, then recognition capability is improved, but response time is delayed and power consumption increases
Solution Approach 1:
The patent extracts the AI recognition model from the server environment and embeds it directly into the mobile terminal. This allows the terminal to perform local inference operations, eliminating network transmission delays and achieving real-time product recognition while maintaining high accuracy.
Solution Approach 2:
The patent introduces an AI model management server as an intermediary that handles model training, version management, and distribution. This separates the heavy computational tasks of model training from the terminal device, allowing the terminal to use pre-trained models locally for fast inference without needing to perform training itself.
2Measurement precision
If high-definition video is transmitted to server in real time, then product recognition accuracy is improved, but data transmission amount increases significantly
Solution Approach 1:
The patent extracts the video processing function from server-based analysis and performs it locally on the terminal using embedded AI models. The terminal processes video frames locally and only transmits minimal recognition results or model update requests to the server, dramatically reducing data transmission volume.
Solution Approach 2:
The patent uses pre-trained AI recognition models that are copied from the server to the terminal device. These models enable the terminal to perform accurate product recognition locally without needing to transmit video data to the server for analysis, thus reducing data transmission requirements.
3Measurement precision
If AI recognition model is updated to recognize new products, then recognition rate for new products is improved, but learning time and computational amount increase
Solution Approach 1:
The patent implements preliminary action by pre-training comprehensive AI recognition models on the server using large datasets of product information. These pre-trained models are then deployed to terminals, enabling them to recognize new products immediately without requiring local training time. When new products are introduced, updated models are distributed to terminals in advance.
Solution Approach 2:
The patent uses an AI model management server as an intermediary that handles all model training and updating operations. This server acts as a centralized platform that trains models using powerful computational resources, then distributes the trained models to multiple terminals. This separation allows terminals to benefit from continuously improved recognition capabilities without bearing the computational burden of model training.
4Adaptability or versatility
If multiple AI recognition models are stored in terminal, then product recognition versatility is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal AI model management architecture that handles multiple recognition models through a unified interface. The system includes model download management, version control, and automatic updates, allowing terminals to support diverse product categories and continuously expand recognition capabilities through a single integrated system rather than separate management mechanisms for each model.
Data Source
AI summary
A mobile artificial intelligence (AI) device includes a camera configured to capture a video of a product, and a neural processing unit (NPU) configured to perform inference with an AI recognition model to recognize product information from the captured video on the mobile AI device. A battery supplies power to the NPU. A transceiver is configured to transmit only the recognized product information to a remote server without transmitting the captured video, and to receive additional commercial information corresponding to the recognized product information. A display is configured to present the additional commercial information.