Sand mining vessel recognition method based on image saliency detection
A sand mining ship, a remarkable technology, applied in the field of computer vision, can solve the problems of inability to effectively detect image noise, inability to effectively integrate global semantic information and local feature information, etc., to achieve the effect of reducing the loss of global semantic information
Inactive Publication Date: 2021-02-02
ZHEJIANG UNIV OF WATER RESOURCES & ELECTRIC POWER
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Aiming at the inability to effectively integrate the global semantic information and local feature information and the inability to effectively detect the noise in the picture in most of the current sand mining ship detection methods, a recursive and decreasing deep fusion semantic information model is proposed.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0024] The present invention will be further described below in conjunction with accompanying drawing.
[0025] The model of the present invention is based on the Caffe deep learning framework.
[0026] Designing an end-to-end multi-layer neural network can directly map the input image into the required pixel-level salient detection map. To this end, (1) first, the model can generate multi-level saliency maps to capture global semantic or local features at different levels. (2) The model needs sufficient depth to capture the specific information of the picture and the hidden contextual comparison information.
[0027] Such as figure 1 As shown, the pixel layer neural network model is designed. The initial picture (the initial picture size is 256×256, unit: pixel) passes through the deep full convolutional neural network, and the corresponding result pictures are generated in different convolutional layers. The depth of these convolutions and neural networks of different siz...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a sand mining ship recognition method based on image saliency detection. The method proposes a new strong supervision for the situation that partial salient objects (sand mining ship hull) detection models appear to lack global semantic information in some pictures. Significance detection method. The model is divided into two layers. The first layer mainly uses the full convolutional neural network to capture the global semantic information and local feature information of the sand mining ship picture at the pixel level, and label the sand mining ship hull. The second layer denoises the saliency map obtained by the first layer layer by layer through the recursive-decreasing model we proposed, and uses local information to supplement the lack of global semantics, and strengthens the boundary features of salient objects. The model performs well on the collected sand dredger dataset and performs well on the existing 6 SOD datasets.
Description
technical field [0001] The invention relates to the field of computer vision, and is mainly aimed at the salience detection of illegal sand mining ships on the river surface. Background technique [0002] In the era of rapid economic development, people's demand for the economy is greater than before. Many people began to mine sand illegally in rivers and resell it. We can use the saliency detection technology to detect such illegal behaviors. Although there are some technologies on the market that can detect illegal sand mining ships, they have the following shortcomings: (1) the object detection of sand mining ships in the river surface picture is not obvious; (2) for the detected sand mining ships, the outline is not clear, And sometimes it is very blurry, and part of the global semantic information is lost; (3) The existing saliency detection model cannot effectively exclude impurity elements in the river surface pictures such as river surface waves or water waves, and...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
Inventor 孙丰马艳娜卢克
Owner ZHEJIANG UNIV OF WATER RESOURCES & ELECTRIC POWER