Multi-Attribute Convolutional Neural Network for Human Surveillance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current human attribute recognition technologies are inefficient in simultaneously recognizing multiple attributes due to the complexity of preprocessing operations and the need for separate attribute recognition models for each attribute.
Innovation Solution
A multi-attribute convolutional neural network model is trained on pre-obtained images to recognize multiple human attributes simultaneously, allowing for real-time recognition of human attributes in surveillance images using a single model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate attribute recognition models are used for each human attribute, then the recognition precision for each attribute can be maintained, but the system complexity and preprocessing operations increase significantly
Solution Approach 1:
The patent combines multiple separate attribute recognition models into a single multi-attribute convolutional neural network model. This unified model simultaneously recognizes multiple human attributes (such as gender, age, clothing type, and color) through integrated training and inference, thereby reducing system complexity while maintaining recognition precision through joint optimization of all attributes
2Adaptability or versatility
If separate attribute recognition models are used for each human attribute, then comprehensive attribute coverage can be achieved, but the processing time and efficiency deteriorate
Solution Approach 1:
The patent implements continuous attribute recognition through a unified multi-attribute convolutional neural network model. The model processes all attributes simultaneously in a single forward pass, eliminating the need for sequential processing of multiple separate models. This continuous action approach maintains comprehensive attribute coverage while significantly improving processing efficiency by removing redundant preprocessing and inference steps
3Adaptability or versatility
If multiple separate attribute recognition models are deployed, then all human attributes can be recognized, but the computational resources and time consumption increase
Solution Approach 1:
The patent merges multiple attribute recognition tasks into a single multi-attribute convolutional neural network model that performs all recognitions simultaneously. This consolidation reduces recognition time by eliminating sequential processing delays and redundant computational steps, while maintaining the capability to recognize all human attributes through integrated feature extraction and classification
Data Source
AI summary
This application discloses a human attribute recognition method performed at a computing device. The method includes: determining a human body region image in a surveillance image; inputting the human body region image into a multi-attribute convolutional neural network model, to obtain, for each of a plurality of human attributes in the human body region image, a probability that the human attribute corresponds to a respective predefined attribute value, the multi-attribute convolutional neural network model being obtained by performing multi-attribute recognition and training on a set of pre-obtained training images by using a multi-attribute convolutional neural network; determining, for each of the plurality of human attributes in the human body region image, the attribute value of the human attribute based on the corresponding probability; and displaying the attribute values of the plurality of human attributes next to the human body region image.