Concrete Dam Defect Recognition in Time-Sequence Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately and efficiently recognize defects in time sequence images of concrete dams due to the presence of numerous background frames and the need for direct video processing, which is hindered by video compression and encoding, affecting defect detection and structural safety.
Innovation Solution
An intelligent recognition method using a two-stream network with a time-dimensional self-attention mechanism, combined with a Transformer network, to extract global feature relations, and an objective function based on distance intersection-over-union to enhance defect location accuracy, along with a convolutional network for defect type recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If direct video processing is used for defect detection, then comprehensive defect information can be obtained, but computational complexity increases and processing efficiency decreases due to video compression and encoding requirements
Solution Approach 1:
The patent extracts only the necessary defect-related information from video data by converting videos into time-dimensional image sequences and using a two-stream network to extract spatial and temporal features. This extraction approach obtains comprehensive defect information while avoiding the computational burden of processing entire compressed video streams, thus improving both detection accuracy and processing efficiency.
2Reliability
If the entire time sequence image is processed for defect recognition, then complete defect information is obtained, but processing time increases due to the presence of numerous background frames
Solution Approach 1:
The patent extracts only the relevant defect portions from the time sequence image by using a two-stream network to identify and focus on defect-containing frames while filtering out background frames. This selective extraction maintains complete defect information for accurate recognition while significantly reducing processing time by excluding irrelevant background data.
Solution Approach 2:
The patent performs preliminary filtering of the time sequence image to identify and separate defect-containing frames from background frames before detailed defect recognition. This preliminary action eliminates numerous background frames in advance, reducing the amount of data requiring intensive processing while ensuring no defect information is lost.
3Measurement precision
If a complex model structure is used to capture global feature relations, then defect location accuracy improves, but model training time and computational resources increase
Solution Approach 1:
The patent segments the feature extraction process into two independent streams: a spatial stream for extracting spatial features and a temporal stream for extracting temporal features. This segmentation allows the model to capture global feature relations through coordinated attention mechanisms while training each stream separately, reducing overall training time and computational resource requirements compared to a monolithic complex model.
Data Source
AI summary
Disclosed is an intelligent recognition method for a time sequence image of a concrete dam defect. The method includes: extracting a feature sequence of the time sequence image containing the concrete dam defect; matching a located defect with a real defect by using an objective function; adding a loss term based on a tight sensing intersection-over-union to a loss function of a model so as to pay attention to integrity of a defect sequence and improve accuracy; and extracting a defect feature and recognizing a defect type after completing defect location. According to the present disclosure, the time sequence image of the concrete dam defect is detected effectively, so that a defect in a long image sequence can be located and the defect type can be recognized accurately.