CN114241377A - Ship target detection method, device, equipment and medium based on improved YOLOX - Google Patents
Ship target detection method, device, equipment and medium based on improved YOLOX Download PDFInfo
- Publication number
- CN114241377A CN114241377A CN202111543482.1A CN202111543482A CN114241377A CN 114241377 A CN114241377 A CN 114241377A CN 202111543482 A CN202111543482 A CN 202111543482A CN 114241377 A CN114241377 A CN 114241377A
- Authority
- CN
- China
- Prior art keywords
- yolox
- ship
- network
- improved
- sample data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a ship target detection method, a device, equipment and a medium based on improved YOLOX, wherein the method obtains a sample data set, the sample data set comprises a ship marked image and an unmarked image, and the sample data set is divided into a training set and a test set; constructing a YOLOX network, optimizing a backbone network of the YOLOX network, and replacing the backbone network with ResNet50 to extract small target features of pictures; inputting the sample data set into the YOLOX network for training and testing to obtain a ship detection model; and inputting the image to be detected into the trained ship detection model for detection so as to output a ship detection result. According to the invention, the traditional YOLO algorithm is improved through methods such as network structure improvement, model simplification and the like, so that the improved algorithm is better suitable for a ship target detection task, and ship inspectors are assisted to find potential dangers and make correct judgment as soon as possible.
Description
Technical Field
The invention relates to the technical field of image detection, in particular to a ship target detection method, a device, equipment and a medium based on improved YOLOX.
Background
In recent years, the development of ship systems is being advanced toward intellectualization, unmanned and systematization. Smart ships have become a new hotspot in the international sea world. The intelligent ship combines new technologies such as artificial intelligence, big data and cloud computing, and has the characteristics of safety, environmental protection, energy conservation and high efficiency. Ships are often exposed to fog, high humidity and various sea conditions while underway. At the same time, they must comply with the rules of sea navigation, and these difficulties place higher demands on the perceived ability of the vessel. Therefore, the sensing system of the smart ship is a key part connected with the outside. The coastline of China is long, the ocean supervision tasks are multiple, the problems of illegal fishing, violation of sea areas and the like bring hidden dangers to ocean safety, and therefore real-time and rapid detection of ship targets is very necessary.
For marine target detection, conventional detection methods are mainly based on images and radar. An embedded visual system is constructed by Mohammerid, Zabidi and the like, an enhancement and smoothing method is adopted for a water surface target which is possibly influenced by sunlight reflection, Hu invariant moment characteristics of four ships are extracted, and different neural networks are utilized to identify the ships. The predecessor also proposed a moving object detection method combining background and interframe difference methods. However, the traditional method has low detection rate and poor generalization capability in extremely complex marine environments; therefore, research on fast real-time algorithms, high accuracy and high reliability has become a hotspot. Deep learning based offshore target detection has solved this problem well. Compared with the traditional identification method, the deep learning has stronger characteristic expression capability on the target.
In recent years, convolutional neural networks have made great progress in the field of target detection. The problem is well solved by marine target detection based on deep learning, and compared with the traditional identification method, the deep learning has stronger characteristic expression capability on the target. There are two main types: the region-based Ross R-cnn series algorithm proposed by girshick et al. The regression idea based YOLO series algorithm proposed by Redmon et al. R-CNN first generates candidate regions, then inputs them into the CNN convolutional neural network, performs feature extraction, compares the features with sample features, and then determines the location of the target region. The algorithm is high in detection accuracy, but the V3 algorithm with low calculation speed does not generate a candidate box, and the detection is directly converted into a regression problem. The target position can be obtained only by one-time detection, so that the detection speed is improved, and the real-time detection requirement is met. During actual inspection, the vessel is often far from the shoreline, and there are many small-sized images of the vessel on the sea surface. The current YOLO series algorithm has poor small-scale application effect and has a missing detection phenomenon.
Disclosure of Invention
In order to solve the technical problems, the invention provides a ship target detection method, a device, equipment and a medium based on improved YOLOX (YOLOX), which are used for solving the problems that dense small-size ships are difficult to identify, an anchor-free implementation method of a YOLO algorithm is difficult, and the traditional YOLO algorithm is improved by methods of network structure improvement, model simplification and the like aiming at the problems that the ship target detection process is easy to be interfered by the outside, the small target detection effect is poor and the small target false identification rate is high, so that the improved algorithm is better suitable for a ship target detection task, and ship inspectors are assisted to find potential dangers and make correct judgment early.
In order to achieve the purpose, the technical scheme of the invention is as follows:
the ship target detection method based on the improved YOLOX comprises the following steps:
acquiring a sample data set, wherein the sample data set comprises a ship marked image and an unmarked image, and is divided into a training set and a test set;
constructing a YOLOX network, optimizing a backbone network of the YOLOX network, and replacing the backbone network with ResNet50 to extract small target features of pictures;
inputting the sample data set into the YOLOX network for training and testing to obtain a ship detection model;
and inputting the image to be detected into the trained ship detection model for detection so as to output a ship detection result.
Preferably, the method further comprises the following steps: and (3) increasing the sample data set by adopting CUTMIX, and then carrying out network training and testing, thereby improving the extraction effect of the network model on the image characteristics.
Preferably, when a training set is input into the YOLOX network for training, the Class is set to 1, the detected targets are classified into two types, namely ship and background, and the number of targets with the largest pictures is set to 10.
Preferably, the output of the YOLOX network is connected to a CFE module, which comprises two branches, wherein the left branch comprises in sequence: convolution layers with convolution kernel sizes of 1x k and k x1, and performing convolution and operation on input data respectively and simultaneously by utilizing convolution of 1x k and k x 1; the right branch comprises in sequence: the convolution kernel is two convolution layers with the sizes of k 1 and 1 k, and input data are simultaneously subjected to convolution and operation by utilizing the convolution of k 1 and 1 k respectively.
Preferably, the sample data set is arranged in a VOC data set.
Preferably, the sample data set and the image to be recognized are preprocessed, and the preprocessing comprises random horizontal or vertical turning, cutting and scale transformation.
Preferably, the backbone network of the YOLOX network is replaced by Hourglass.
A ship target detection device based on improved YOLOX comprises: an acquisition module, an image processing module, a model acquisition module and a recognition module, wherein,
the acquisition module is used for acquiring a sample data set, wherein the sample data set comprises a ship marked image and an unmarked image and is divided into a training set and a test set;
the image processing module is used for preprocessing the sample data set;
the model acquisition module is used for constructing a YOLOX network, and a backbone network of the YOLOX network is replaced by ResNet50 to extract small target features of pictures and input sample data sets into the YOLOX network for training and testing to obtain a ship detection model;
and the recognition module is used for inputting the image to be detected into the trained ship detection model for detection and outputting a ship detection result.
A computer device, comprising: a memory for storing a computer program; a processor for implementing the improved YOLOX based ship target detection method as described in any one of the above when the computer program is executed.
A readable storage medium having stored thereon a computer program which, when executed by a processor, implements the improved YOLOX-based ship target detection method as defined in any one of the above.
Based on the technical scheme, the invention has the beneficial effects that: with the continuous development of artificial intelligence technology, deep learning methods begin to permeate into various fields, and intelligent supervision of oceans is urgently needed for people, China possesses abundant ocean resources and wide sea areas, and ocean economy is one of the main supporting points for human development in the 21 st century. In the invention, a latest detection technology YOLOX is used, the detection precision and speed of natural images reach the highest level at present, on the basis, a network structure is improved, strategy deletion is carried out on the YOLOX, a data set is changed, the YOLOX is more suitable for ship detection in the sea, the YOLOX is easier to deploy, the method can be used for monitoring port and channel, and intelligent monitoring on fishing ships and cargo ships is facilitated. Meanwhile, the method has wide application prospects in the aspects of port management, cross-border ship detection, autonomous ships, safe navigation and the like, and in practical application, the algorithm is better balanced than a YOLO v3 network, a YOLO v4 network and a YOLO v5 network.
Drawings
FIG. 1 is a diagram of an application environment of a ship target detection method based on improved YOLOX in one embodiment;
FIG. 2 is a flow chart of a ship target detection method based on improved YOLOX in one embodiment;
FIG. 3 is a model framework diagram in an embodiment of a ship target detection method based on improved YOLOX;
FIG. 4 is a schematic structural diagram of a ship target detection device based on improved YOLOX in one embodiment;
FIG. 5 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
The ship target detection method based on the improved YOLOX provided by the embodiment of the application can be applied to the application environment shown in FIG. 1. As shown in FIG. 1, the application environment includes a computer device 110. The computer device 110 may obtain a sample data set; the computer device 110 may construct a YOLOX network, optimize a backbone network of the YOLOX network, and replace the backbone network with ResNet50 for extracting small target features of a picture; the computer device 110 may input the sample data set into the YOLOX network for training and testing to obtain a ship detection model; the computer device 110 may input the image to be detected into the trained ship detection model for detection, so as to output a ship detection result.
As shown in fig. 2, a ship target detection method based on improved YOLOX is provided, which specifically includes the following steps:
step 210, a sample data set is obtained, wherein the sample data set comprises a ship marked image and an unmarked image, and the sample data set is divided into a training set and a test set.
In this embodiment, a YOLOX-Darknet53 network is used as a backbone network of the detector, and since there is a poor small target detection effect in the YOLOV3 algorithm, the speed is slightly behind compared with the current one-stage detector, and there is still a space for improving the model performance when introduced into YOLOX, for this problem, as shown in FIG. 3, we use ResNet50 to replace the backbone network therein, so as to improve the accuracy of the model detection. And the ResNet series has great optimization strength, and can well meet the requirements of actual deployment and the like.
The data reduction and increase strategy in the YOLOX-Darknet53 network was adjusted to CUTMIX. The method has the advantages that the data enhancement strategy of CUTMIX is adopted to increase the sample data set and then carry out network training and testing, the extraction effect of the network model on image features is improved, non-information pixels can not appear in the training process of the strategy, accordingly, the training efficiency can be improved, objects can be identified from local views through the requirement of the model, the positioning capacity of the model can be further enhanced by adding information of other samples in the cut region, and the method is as follows:
xAand xBAre two different training samples, yAAnd yBIs the corresponding label value, what cpu mix needs to generate is a new training sample and corresponding label:
and
the formula is as follows:
M∈{0，1}W×Hin order for dropd to drop a partial region and fill the binary mask,
is a pixel-by-pixel multiplication, l is a binary mask with all elements 1, λ belongs to the Beta distribution as well as the Mixup: λ ∈ Beta (α, α), and let α ═ 1 then λ obeys a uniform distribution of (0, 1).
To sample the binary mask M, the bounding box B of the cropped region is first divided into (r)x,ry,rw,rh) Sampling is performed to obtain a sample xAAnd xBAnd performing indication and calibration on the cutting area. The rectangular mask M is sampled (length and width proportional to sample size).
The bounding box sampling formula for the cropped regions is as follows:
where W is the bounding box width, H is the bounding box height, and Unif () is the same function. Ensuring that the cutting area has a ratio of
After the trimming area B is determined, the trimming area B in the mask M is set to 0, and the other areas are set to 1. The sampling of the mask is completed and then the cropping zone B in sample a is removed, cropped and then filled into sample a.
And step 230, inputting the sample data set into the YOLOX network for training and testing to obtain a ship detection model.
In this embodiment, set up training set classification figure, because to the marine ship detection of ocean, the classification problem should be simplified, then we will set up in the training set Class 1, divide into boats and ships and background two types with the target that detects, only keep the image of boats and ships mark in the data set simultaneously, increase Batch simultaneously, can promote model training efficiency.
Since the ship volume is large and the number of ships in the image is generally small, we set the detection number to 10, and max _ labels represents the target number with the largest number of pictures. Since our dataset mimics the layout of the VOC dataset files, without year age information, this element of year will be deleted.
And 240, inputting the image to be detected into the trained ship detection model for detection so as to output a ship detection result.
In this example we will set the end-to-end YOLOX, adding two extra conv layers in the model, one-to-one label assignment and stop gradient. This enables the detector to perform in an end-to-end manner, but slightly reduces performance and inference speed, so i improve the two conv layers, and we replace the two conv layers with CFE modules, which use the CFE modules in the CFEnet network for reference, replace the original 1x1conv2d structure, enrich information extraction, and obtain better results than the original algorithm. The CFE module is intended to enhance the shallow nature of SSD detection of small targets. Its sensitivity comes from modules such as inclusion, Xception, Large separation and ResNeXt. This CFE module includes two branches, and wherein, left branch includes in proper order: convolution layers with convolution kernel sizes of 1x k and k x1, and performing convolution and operation on input data respectively and simultaneously by utilizing convolution of 1x k and k x 1; the right branch comprises in sequence: the convolution kernel is two convolution layers with the sizes of k 1 and 1 k, and input data are simultaneously subjected to convolution and operation by utilizing the convolution of k 1 and 1 k respectively.
In one embodiment, for ResNet50, we can also replace the network with Hourglass as the backbone network for feature extraction. Hourglass is mainly used for attitude detection at present, and the design of the network is mainly due to the requirement of grabbing information of each scale, for example, some local information is important for identifying some features (such as faces, hands and the like), so that the ship detection is conjointly realized, and for ships, the local information thereof is necessary for identifying the shape features of the ship bottom. The identification target needs to have a good understanding of the whole target, so that a lot of local characteristic information is grasped and combined, and for a ship, the ship can be well identified by grasping the characteristic of the ship bottom. Meanwhile, the hourglass is a simple and minimized design, has the capability of capturing all characteristic information and making final pixel level prediction, and the network of the step can realize a method for completing an anchor-free. For propagation detection, the detected target type is single, and a data enhancement module of the YOLOX can be cut off, so that the accuracy of a trained model is not reduced, the training efficiency is improved, and the deployment of the YOLOX in ship detection is completed more quickly.
In one embodiment, as shown in fig. 4, there is provided a ship target detection apparatus 300 based on improved YOLOX, including: an acquisition module 310, an image processing module 320, a model acquisition module 330, and a recognition module 340, wherein,
the obtaining module 310 is configured to obtain a sample data set, where the sample data set includes a vessel labeled image and an unlabeled image, and the sample data set is divided into a training set and a test set;
the image processing module 320 is configured to pre-process the sample data set;
the model obtaining module 330 is configured to construct a YOLOX network, where a backbone network of the YOLOX network is replaced with ResNet50, so as to extract small target features of a picture, and to input a sample data set into the YOLOX network for training and testing to obtain a ship detection model;
the recognition module 340 is configured to input the image to be detected into the trained ship detection model for detection, so as to output a ship detection result.
The apparatuses or modules illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 5. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method for improved YOLOX based ship target detection. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 5 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a readable storage medium is provided, having stored thereon a computer program which, when being executed by a processor, implements the improved YOLOX-based ship target detection method as defined in any one of the above.
It will be understood by those skilled in the art that all or part of the processes of the ship target detection method based on improved YOLOX according to the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the computer program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The above embodiments are merely preferred examples of the present application, and are not intended to limit the present application, and those skilled in the art may make various modifications and changes. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the embodiments of the present application shall be included in the protection scope of the embodiments of the present application.
Claims (10)
1. The ship target detection method based on the improved YOLOX is characterized by comprising the following steps:
acquiring a sample data set, wherein the sample data set comprises a ship marked image and an unmarked image, and is divided into a training set and a test set;
constructing a YOLOX network, optimizing a backbone network of the YOLOX network, and replacing the backbone network with ResNet50 to extract small target features of pictures;
inputting the sample data set into the YOLOX network for training and testing to obtain a ship detection model;
and inputting the image to be detected into the trained ship detection model for detection so as to output a ship detection result.
2. The improved YOLOX-based ship target detection method according to claim 1, further comprising the steps of:
and (3) increasing the sample data set by adopting CUTMIX, and then carrying out network training and testing, thereby improving the extraction effect of the network model on the image characteristics.
3. The improved YOLOX-based ship target detection method as claimed in claim 2, wherein when a training set is input into the YOLOX network for training, the Class is set to 1, the detected targets are classified into two classes of ship and background, and the number of targets with the largest number of pictures is set to 10.
4. The improved YOLOX-based ship target detection method as claimed in claim 1, wherein the output of the YOLOX network is connected to a CFE module, the CFE module comprises two branches, wherein the left branch comprises in sequence: convolution layers with convolution kernel sizes of 1x k and k x1, and performing convolution and operation on input data respectively and simultaneously by utilizing convolution of 1x k and k x 1; the right branch comprises in sequence: the convolution kernel is two convolution layers with the sizes of k 1 and 1 k, and input data are simultaneously subjected to convolution and operation by utilizing the convolution of k 1 and 1 k respectively.
5. The improved YOLOX-based ship target detection method as claimed in claim 1, wherein the sample data set is arranged in a VOC data set.
6. The improved YOLOX-based ship target detection method as claimed in claim 1, wherein the sample data set and the image to be identified are preprocessed, the preprocessing includes random horizontal or vertical flipping, cropping, and scaling.
7. The improved YOLOX-based ship target detection method as claimed in claim 1, wherein the backbone network of the YOLOX network is replaced by Hourglass.
8. A ship target detection device based on improved YOLOX is characterized by comprising: an acquisition module, an image processing module, a model acquisition module and a recognition module, wherein,
the acquisition module is used for acquiring a sample data set, wherein the sample data set comprises a ship marked image and an unmarked image and is divided into a training set and a test set;
the image processing module is used for preprocessing the sample data set;
the model acquisition module is used for constructing a YOLOX network, and a backbone network of the YOLOX network is replaced by ResNet50 to extract small target features of pictures and input sample data sets into the YOLOX network for training and testing to obtain a ship detection model;
and the recognition module is used for inputting the image to be detected into the trained ship detection model for detection and outputting a ship detection result.
9. A computer device, comprising: a memory for storing a computer program; a processor for implementing the improved YOLOX based ship target detection method as claimed in any one of claims 1 to 7 when executing the computer program.
10. A readable storage medium, characterized in that the readable storage medium has stored thereon a computer program which, when being executed by a processor, implements the improved YOLOX-based ship target detection method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111543482.1A CN114241377A (en) | 2021-12-16 | 2021-12-16 | Ship target detection method, device, equipment and medium based on improved YOLOX |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111543482.1A CN114241377A (en) | 2021-12-16 | 2021-12-16 | Ship target detection method, device, equipment and medium based on improved YOLOX |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114241377A true CN114241377A (en) | 2022-03-25 |
Family
ID=80757297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111543482.1A Pending CN114241377A (en) | 2021-12-16 | 2021-12-16 | Ship target detection method, device, equipment and medium based on improved YOLOX |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114241377A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114463640A (en) * | 2022-04-08 | 2022-05-10 | 武汉理工大学 | Multi-view ship identity recognition method with local feature fusion |
| CN114882556A (en) * | 2022-04-26 | 2022-08-09 | 西北大学 | Method for detecting makeup face of opera character based on improved YooloX |
| CN115187950A (en) * | 2022-09-13 | 2022-10-14 | 安徽中科星驰自动驾驶技术有限责任公司 | Novel balance mask secondary sampling method for deep learning training data enhancement |
| CN115861365A (en) * | 2022-10-11 | 2023-03-28 | 海南大学 | Moving object detection method, system, computer device and storage medium |
| CN116665088A (en) * | 2023-05-06 | 2023-08-29 | 海南大学 | Ship identification and detection method, device, equipment and medium |
-
2021
- 2021-12-16 CN CN202111543482.1A patent/CN114241377A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114463640A (en) * | 2022-04-08 | 2022-05-10 | 武汉理工大学 | Multi-view ship identity recognition method with local feature fusion |
| CN114882556A (en) * | 2022-04-26 | 2022-08-09 | 西北大学 | Method for detecting makeup face of opera character based on improved YooloX |
| CN114882556B (en) * | 2022-04-26 | 2024-03-15 | 西北大学 | Improved YoloX-based drama character dressing face detection method |
| CN115187950A (en) * | 2022-09-13 | 2022-10-14 | 安徽中科星驰自动驾驶技术有限责任公司 | Novel balance mask secondary sampling method for deep learning training data enhancement |
| CN115187950B (en) * | 2022-09-13 | 2022-11-22 | 安徽中科星驰自动驾驶技术有限责任公司 | Novel balance mask secondary sampling method for deep learning image data enhancement |
| CN115861365A (en) * | 2022-10-11 | 2023-03-28 | 海南大学 | Moving object detection method, system, computer device and storage medium |
| CN115861365B (en) * | 2022-10-11 | 2023-08-15 | 海南大学 | Moving object detection method, system, computer device and storage medium |
| CN116665088A (en) * | 2023-05-06 | 2023-08-29 | 海南大学 | Ship identification and detection method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cheng et al. | FusionNet: Edge aware deep convolutional networks for semantic segmentation of remote sensing harbor images | |
| CN114241377A (en) | Ship target detection method, device, equipment and medium based on improved YOLOX | |
| Li et al. | Object detection using convolutional neural networks in a coarse-to-fine manner | |
| Teng et al. | Global to local: Clip-LSTM-based object detection from remote sensing images | |
| Sun et al. | Global Mask R-CNN for marine ship instance segmentation | |
| Zheng et al. | A lightweight ship target detection model based on improved YOLOv5s algorithm | |
| Li et al. | Ship detection of optical remote sensing image in multiple scenes | |
| CN113706481A (en) | Sperm quality detection method, sperm quality detection device, computer equipment and storage medium | |
| Fan et al. | A novel sonar target detection and classification algorithm | |
| Chang et al. | Yolov3 based ship detection in visible and infrared images | |
| Yildirim et al. | Ship detection in optical remote sensing images using YOLOv4 and Tiny YOLOv4 | |
| CN114565824A (en) | Single-stage rotating ship detection method based on full convolution network | |
| CN114529821A (en) | Offshore wind power safety monitoring and early warning method based on machine vision | |
| Zhang et al. | JF-YOLO: the jellyfish bloom detector based on deep learning | |
| Zou et al. | Maritime target detection of intelligent ship based on faster R-CNN | |
| Yin et al. | High-order spatial interactions enhanced lightweight model for optical remote sensing image-based small ship detection | |
| Zhang et al. | Swin-PAFF: A SAR Ship Detection Network with Contextual Cross-Information Fusion. | |
| Huang et al. | A deep learning approach to detecting ships from high-resolution aerial remote sensing images | |
| Zhai et al. | Ship detection based on faster R-CNN network in optical remote sensing images | |
| Shishkin et al. | Implementation of yolov5 for detection and classification of microplastics and microorganisms in marine environment | |
| Lyu et al. | Distinguishing text/non-text natural images with multi-dimensional recurrent neural networks | |
| Li et al. | Research on ROI algorithm of ship image based on improved YOLO | |
| Hu et al. | Two-stage insulator self-explosion defect detection method based on Mask R-CNN | |
| Wang et al. | Accurate segmentation of Ulva prolifera regions with superpixel and CNNs | |
| Liu et al. | A modified-yolov5s model for defect detection of printed circuit board |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |