CN113468363A - Target detection acceleration method for remote sensing image - Google Patents
Target detection acceleration method for remote sensing image Download PDFInfo
- Publication number
- CN113468363A CN113468363A CN202110755017.8A CN202110755017A CN113468363A CN 113468363 A CN113468363 A CN 113468363A CN 202110755017 A CN202110755017 A CN 202110755017A CN 113468363 A CN113468363 A CN 113468363A
- Authority
- CN
- China
- Prior art keywords
- input
- region
- value
- determining
- cache
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 230000001133 acceleration Effects 0.000 title claims description 13
- 238000004364 calculation method Methods 0.000 claims abstract description 102
- 238000012545 processing Methods 0.000 claims abstract description 75
- 239000010410 layer Substances 0.000 claims description 26
- 239000002356 single layer Substances 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 6
- 238000009826 distribution Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 8
- 238000005457 optimization Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The present disclosure provides a method for accelerating target detection of a remote sensing image, which includes: acquiring input data of a remote sensing image; segmenting the input data to obtain at least one input subarea; calculating a hash value of the input sub-region; searching a corresponding value of the hash value in the input subarea in a cache according to the hash value; and determining a processing task, and processing the input sub-area according to the processing task to obtain output data. According to the method for accelerating the target detection of the remote sensing image, the input subarea is processed according to the processing task, so that the effect of saving redundant calculation overhead is achieved, and the processing speed of the target detection model of the remote sensing image is improved.
Description
Technical Field
The disclosure relates to the field of processing of remote sensing images of neural networks, in particular to a target detection acceleration method for a remote sensing image.
Background
Remote sensing images are widely applied to different scenes of various industries such as agriculture, archaeology, geology, city planning and the like at present. However, the size of the remote sensing image is far larger than that of a common image, and the number and the spatial imbalance of the target are presented in the image, so that the processing speed of target detection of the remote sensing image becomes a great bottleneck, and the improvement of the target detection speed of the remote sensing image becomes a problem to be solved urgently.
For the classical target detection model, a lot of optimization work is generated, so that the processing speed can achieve the real-time effect. For remote sensing images, the existing method still has certain limitations: (1) the classical model optimization method does not utilize the characteristics of remote sensing images, a plurality of optimization spaces are not effectively utilized, and the purpose of optimization cannot be achieved to the maximum extent; (2) the optimization method aiming at a specific target has excellent effect on the specific target, but has no effect on other targets, and lacks of general applicability; (3) although the filter based on network training can well remove the remote sensing image block which does not contain the target object, a large amount of time is consumed for training, and the brought additional cost is large.
Disclosure of Invention
Technical problem to be solved
The present disclosure provides a method for accelerating target detection of a remote sensing image, so as to at least solve the problems in the prior art.
(II) technical scheme
In order to achieve the above object, the present disclosure provides a method for accelerating target detection of a remote sensing image, including:
acquiring input data of a remote sensing image;
segmenting the input data to obtain at least one input subarea;
calculating a hash value of the input sub-region;
searching a corresponding value of the hash value in the input subarea in a cache according to the hash value;
and determining a processing task, and processing the input sub-area according to the processing task to obtain output data.
In some embodiments of the present disclosure, the calculating the hash value of the input sub-region comprises:
flattening the input sub-region into a one-dimensional vector, calculating a hash value of the one-dimensional vector based on a stochastic projected locality sensitive hash function,
the formula of the locality sensitive hash function h (x) is as follows:
wherein x is represented as the one-dimensional vector and v is represented as a random vector satisfying a standard gaussian distribution.
In some embodiments of the present disclosure, the looking up, in a cache, a corresponding value of the hash value in the input subregion according to the hash value includes:
forming a one-dimensional binary vector by the calculated values of the N locality sensitive hash functions, wherein N is an integer greater than or equal to 1;
and regarding the one-dimensional binary vector as a binary number, and taking an integer value corresponding to the binary number as a key value to search a corresponding value of the hash value in the input sub-area.
In some embodiments of the present disclosure, the determining a processing task, and processing the input sub-region according to the processing task to obtain output data includes:
determining that the processing task is a filtering task, and only storing the key value in a cache;
and determining that the corresponding value exists in the cache, and directly taking the empty result as output data.
In some embodiments of the present disclosure, the determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
determining that the processing task is a filtering task, and only storing the key value in a cache;
determining that the corresponding value does not exist in the cache, inputting the input sub-region into a corresponding calculation module for calculation, and outputting a calculation result, wherein the calculation module is a network layer for inputting vertex coordinates of a rectangular frame for calibrating a target region to be detected, the calculation module comprises a single-layer calculation module consisting of a single network layer or a plurality of calculation modules consisting of continuous network layers, and the calculation result is the vertex coordinates of the rectangular frame for marking the target to be detected;
and determining that the calculation result is empty, and storing the hash value of the input sub-region into the cache, wherein the cache is in a hash table structure.
In some embodiments of the present disclosure, the determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
determining that the processing task is a reuse task, and storing a calculation result of the input sub-region after calculation by a calculation module in the cache;
and determining that the corresponding value exists in the cache, and directly taking a calculation result of the input sub-region stored in the cache after calculation by a calculation module as output data, wherein the calculation module comprises a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers.
In some embodiments of the present disclosure, the determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
determining that the processing task is a reuse task, and storing a calculation result of the input sub-region after calculation by a calculation module in the cache;
determining that the corresponding value does not exist in the cache, inputting the input sub-region into a corresponding calculation module for calculation and outputting a calculation result, wherein the calculation module is a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers;
and determining that the calculation result is not empty, taking the hash value of the input sub-region as a key value, and taking the calculation result as a storage value to be added into the cache.
In some embodiments of the present disclosure, further comprising:
and arranging the calculation results according to the relative position of the input sub-area in the original input, wherein the relative position in the original input refers to the position corresponding to the input sub-area.
In some embodiments of the present disclosure, the segmenting the input data into at least one input sub-region includes:
representing the size of the input sub-region as a sub-value;
representing the size of the original input as an original value;
dividing the input data or filling 0 element in the input data to make the original value be integral multiple of the sub-value;
wherein each input sub-region is a square region with the same size.
(III) advantageous effects
According to the technical scheme, the remote sensing image target detection acceleration method disclosed by the invention at least has one or part of the following beneficial effects:
(1) according to the method for accelerating the target detection of the remote sensing image, the input subarea is processed according to the processing task, so that the effect of saving redundant calculation overhead is achieved, and the processing speed of the target detection model of the remote sensing image is improved.
(2) According to the remote sensing image target detection acceleration method, the Hash value of the input data is used as the searching key value of the cache, a large amount of input data which do not contain the target object can be removed, the calculation result in the network layer is reused by detecting the similarity of the target object, the method has the advantages of universality and plug-and-play, and is suitable for various application scenes of remote sensing image target detection.
Drawings
FIG. 1 is a schematic diagram of a method for accelerating target detection in a remote sensing image according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of the substeps of operation S4 in FIG. 1;
fig. 3 is a schematic diagram of the sub-steps of operation S2 in fig. 1.
Detailed Description
The present disclosure provides a method for accelerating target detection of a remote sensing image, which includes: acquiring input data of a remote sensing image; dividing input data to obtain at least one input subarea; calculating a hash value of the input sub-region; searching a corresponding value of the hash value in the input subarea in the cache according to the hash value; and determining a processing task, and processing the input sub-area according to the processing task to obtain output data. The method can remove a large amount of input data which do not contain the target object, and reuses the calculation result in the network layer by detecting the similarity of the target object; the effect of saving redundant calculation overhead is achieved, and the target detection speed of the remote sensing image is further improved.
For the purpose of promoting a better understanding of the objects, aspects and advantages of the present disclosure, reference is made to the following detailed description taken in conjunction with the accompanying drawings. This disclosure may, however, be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. In the drawings, the size and relative sizes of layers and regions may be exaggerated for clarity, and like reference numerals designate like elements throughout.
Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.
All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.
Where a convention analogous to "at least one of A, B and C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B and C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.). Where a convention analogous to "A, B or at least one of C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B or C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).
The present disclosure provides a method for accelerating target detection in remote sensing images, as shown in fig. 1, the method includes operations S1 to S5:
operation S1: acquiring input data of a remote sensing image;
operation S2: dividing input data to obtain at least one input subarea;
operation S3: calculating a hash value of the input sub-region;
operation S4: searching a corresponding value of the hash value in the input subarea in the cache according to the hash value;
operation S5: and determining a processing task, and processing the input sub-area according to the processing task to obtain output data.
The remote sensing image target detection acceleration method processes the input subarea according to the processing task, so that the effect of saving redundant calculation overhead is achieved, and the processing speed of the remote sensing image target detection model is improved.
In operation S3, the calculating the hash value of the input sub-region includes:
flattening the input sub-region into a one-dimensional vector, calculating the hash value of the one-dimensional vector based on the locality sensitive hash function of random projection,
the formula for the locality sensitive hash function h (x) is:
where x is represented as a one-dimensional vector and v is represented as a random vector satisfying a standard gaussian distribution.
In operation S4, as shown in fig. 2, finding the corresponding value of the hash value in the input sub-region in the cache according to the hash value includes operations S41 and S42 as follows:
operation S41: and combining the calculated values of the N locality sensitive hash functions into a one-dimensional binary vector, wherein N is an integer greater than or equal to 1.
Operation S42: and taking the one-dimensional binary vector as a binary number, and taking an integer value corresponding to the binary number as a key value to search a corresponding value of the hash value in the input sub-area.
The hash value of the input data is used as the cached searching key value, a large amount of input data without the target object can be removed, the calculation result in the network layer is reused by detecting the similarity of the target object, and the method has the advantages of universality and plug-and-play, and is suitable for various application scenes of remote sensing image target detection.
In operation S5, determining a processing task, and processing the input sub-region according to the processing task to obtain output data includes:
and determining that the processing task is a filtering task, and only storing the key value in the cache.
And determining that a corresponding value exists in the cache, and directly taking the empty result as output data. And if the output data is a null result, the result indicates that no effective target exists in the cache.
As another embodiment, in operation S5, determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
and determining that the processing task is a filtering task, and only storing the key value in the cache.
Determining that no corresponding value exists in the cache, inputting the input sub-region into a corresponding calculation module for calculation and outputting a calculation result, wherein the calculation module is a network layer for inputting vertex coordinates of a rectangular frame for calibrating a target region to be detected, the calculation module is a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers, and the calculation result is the vertex coordinates of the rectangular frame for marking the target to be detected;
and determining that the calculation result is null, and storing the hash value of the input sub-region into a cache, wherein the cache is in a hash table structure.
As another embodiment, in operation S5, determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
and determining the processing task as a reuse task, and storing a calculation result of the input sub-region after calculation by the calculation module in the cache.
And determining that the corresponding value exists in the cache, and directly taking a calculation result of the input sub-region after calculation by the calculation module as output data, wherein the calculation module comprises a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers.
As another embodiment, in operation S5, determining a processing task, and processing the input sub-region according to the processing task to obtain output data, further includes:
and determining the processing task as a reuse task, and storing a calculation result of the input sub-region after calculation by the calculation module in the cache.
And determining that no corresponding value exists in the cache, inputting the input sub-region into a corresponding calculation module for calculation and outputting a calculation result, wherein the calculation module is a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers.
And determining that the calculation result is not null, using the hash value of the input sub-region as a key value, and using the calculation result as a storage value to be added into the cache.
As another embodiment of the present disclosure, in the remote sensing image target detection acceleration method of the present disclosure, operation S5: determining a processing task, and after processing the input sub-region according to the processing task to obtain output data, the method further comprises the following steps:
and arranging the calculation results according to the relative position of the input subarea in the original input, wherein the relative position in the original input refers to the position corresponding to the input subarea, and the calculation results comprise the calculation results which are hit in the cache and taken out of the cache, and also comprise the calculation results which are not hit in the cache and calculated by the calculation module.
In operation S2, as shown in fig. 3, the input data is divided into at least one input sub-region, which includes the following operations S21-S23:
s21: the size of the input sub-region is represented as a sub-value.
S22: the size of the original input is expressed as the original value.
S23: the input data is divided or 0 elements are filled in the input data so that the original value is an integral multiple of the sub-value.
Wherein each input sub-region is a square region with the same size.
The input remote sensing images with inconsistent lengths can be filled with 0 or cut to enable the remote sensing images to meet corresponding conditions.
The calculation result in the network layer is reused by detecting the similarity of the target object, and the method has the advantages of universality and plug-and-play because no additional training process is needed and the method does not aim at a specific target to be detected, and is suitable for various application scenes for detecting the remote sensing image target.
It should also be noted that directional terms, such as "upper", "lower", "front", "rear", "left", "right", and the like, used in the embodiments are only directions referring to the drawings, and are not intended to limit the scope of the present disclosure. Throughout the drawings, like elements are represented by like or similar reference numerals. In the event of possible confusion for understanding of the present disclosure, conventional structures or configurations will be omitted, and the shapes and sizes of the components in the drawings do not reflect actual sizes and proportions, but merely illustrate the contents of the embodiments of the present disclosure.
Unless otherwise indicated, the numerical parameters set forth in the specification and attached claims are approximations that can vary depending upon the desired properties sought to be obtained by the present disclosure. In particular, all numbers expressing quantities of ingredients, reaction conditions, and so forth used in the specification and claims are to be understood as being modified in all instances by the term "about". Generally, the expression is meant to encompass variations of ± 10% in some embodiments, 5% in some embodiments, 1% in some embodiments, 0.5% in some embodiments by the specified amount.
The use of ordinal numbers such as "first," "second," "third," etc., in the specification and claims to modify a corresponding element does not by itself connote any ordinal number of the element or any ordering of one element from another or the order of manufacture, and the use of the ordinal numbers is only used to distinguish one element having a certain name from another element having a same name.
In addition, unless steps are specifically described or must occur in sequence, the order of the steps is not limited to that listed above and may be changed or rearranged as desired by the desired design. The embodiments described above may be mixed and matched with each other or with other embodiments based on design and reliability considerations, i.e., technical features in different embodiments may be freely combined to form further embodiments.
The above-mentioned embodiments are intended to illustrate the objects, aspects and advantages of the present disclosure in further detail, and it should be understood that the above-mentioned embodiments are only illustrative of the present disclosure and are not intended to limit the present disclosure, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present disclosure should be included in the scope of the present disclosure.
Claims (9)
1. A remote sensing image target detection acceleration method comprises the following steps:
acquiring input data of a remote sensing image;
segmenting the input data to obtain at least one input subarea;
calculating a hash value of the input sub-region;
searching a corresponding value of the hash value in the input subarea in a cache according to the hash value;
and determining a processing task, and processing the input sub-area according to the processing task to obtain output data.
2. The remote sensing image target detection acceleration method of claim 1, wherein the calculating the hash value of the input subregion comprises:
flattening the input sub-region into a one-dimensional vector, calculating a hash value of the one-dimensional vector based on a stochastic projected locality sensitive hash function,
the formula of the locality sensitive hash function h (x) is as follows:
wherein x is represented as the one-dimensional vector and v is represented as a random vector satisfying a standard gaussian distribution.
3. The remote sensing image target detection acceleration method according to claim 2, wherein the searching for the corresponding value of the hash value in the input subregion in the cache according to the hash value comprises:
forming a one-dimensional binary vector by the calculated values of the N locality sensitive hash functions, wherein N is an integer greater than or equal to 1;
and regarding the one-dimensional binary vector as a binary number, and taking an integer value corresponding to the binary number as a key value to search a corresponding value of the hash value in the input sub-area.
4. The remote sensing image target detection acceleration method according to claim 3, wherein the determining a processing task, processing the input sub-region according to the processing task to obtain output data, comprises:
determining that the processing task is a filtering task, and only storing the key value in a cache;
and determining that the corresponding value exists in the cache, and directly taking the empty result as output data.
5. The method for detecting and accelerating the remote sensing image target according to claim 3, wherein the determining a processing task and processing the input sub-region according to the processing task to obtain output data further comprises:
determining that the processing task is a filtering task, and only storing the key value in a cache;
determining that the corresponding value does not exist in the cache, inputting the input sub-region into a corresponding calculation module for calculation, and outputting a calculation result, wherein the calculation module is a network layer for inputting vertex coordinates of a rectangular frame for calibrating a target region to be detected, the calculation module comprises a single-layer calculation module consisting of a single network layer or a plurality of calculation modules consisting of continuous network layers, and the calculation result is the vertex coordinates of the rectangular frame for marking the target to be detected;
and determining that the calculation result is empty, and storing the hash value of the input sub-region into the cache, wherein the cache is in a hash table structure.
6. The method for detecting and accelerating the remote sensing image target according to claim 1, wherein the determining a processing task and processing the input sub-region according to the processing task to obtain output data further comprises:
determining that the processing task is a reuse task, and storing a calculation result of the input sub-region after calculation by a calculation module in the cache;
and determining that the corresponding value exists in the cache, and directly taking a calculation result of the input sub-region stored in the cache after calculation by a calculation module as output data, wherein the calculation module comprises a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers.
7. The method for detecting and accelerating the remote sensing image target according to claim 1, wherein the determining a processing task and processing the input sub-region according to the processing task to obtain output data further comprises:
determining that the processing task is a reuse task, and storing a calculation result of the input sub-region after calculation by a calculation module in the cache;
determining that the corresponding value does not exist in the cache, inputting the input sub-region into a corresponding calculation module for calculation and outputting a calculation result, wherein the calculation module is a single-layer calculation module consisting of a single network layer or a calculation module consisting of a plurality of continuous network layers;
and determining that the calculation result is not empty, taking the hash value of the input sub-region as a key value, and taking the calculation result as a storage value to be added into the cache.
8. The remote sensing image target detection acceleration method according to claim 7, further comprising:
and arranging the calculation results according to the relative position of the input sub-area in the original input, wherein the relative position in the original input refers to the position corresponding to the input sub-area.
9. The remote sensing image target detection acceleration method of claim 1, wherein the segmenting the input data into at least one input subregion comprises:
representing the size of the input sub-region as a sub-value;
representing the size of the original input as an original value;
dividing the input data or filling 0 element in the input data to make the original value be integral multiple of the sub-value;
wherein each input sub-region is a square region with the same size.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110755017.8A CN113468363B (en) | 2021-07-01 | 2021-07-01 | Remote sensing image target detection acceleration method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110755017.8A CN113468363B (en) | 2021-07-01 | 2021-07-01 | Remote sensing image target detection acceleration method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113468363A true CN113468363A (en) | 2021-10-01 |
CN113468363B CN113468363B (en) | 2024-05-17 |
Family
ID=77877935
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110755017.8A Active CN113468363B (en) | 2021-07-01 | 2021-07-01 | Remote sensing image target detection acceleration method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113468363B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109284741A (en) * | 2018-10-30 | 2019-01-29 | 武汉大学 | A kind of extensive Remote Sensing Image Retrieval method and system based on depth Hash network |
CN109919108A (en) * | 2019-03-11 | 2019-06-21 | 西安电子科技大学 | Remote sensing images fast target detection method based on depth Hash auxiliary network |
US20210099310A1 (en) * | 2019-09-27 | 2021-04-01 | Boe Technology Group Co., Ltd. | Image processing method, image matching method, device and storage medium |
-
2021
- 2021-07-01 CN CN202110755017.8A patent/CN113468363B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109284741A (en) * | 2018-10-30 | 2019-01-29 | 武汉大学 | A kind of extensive Remote Sensing Image Retrieval method and system based on depth Hash network |
CN109919108A (en) * | 2019-03-11 | 2019-06-21 | 西安电子科技大学 | Remote sensing images fast target detection method based on depth Hash auxiliary network |
US20210099310A1 (en) * | 2019-09-27 | 2021-04-01 | Boe Technology Group Co., Ltd. | Image processing method, image matching method, device and storage medium |
Non-Patent Citations (1)
Title |
---|
闯跃龙;张石清;郭文平;赵小明;: "基于感知哈希和自适应搜索的通用对象检测", 光电子・激光, no. 02 * |
Also Published As
Publication number | Publication date |
---|---|
CN113468363B (en) | 2024-05-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109002039B (en) | Obstacle avoidance reminding method, related device and computer readable storage medium | |
US8339399B2 (en) | Declustering point-of-interest icons | |
CN110990516B (en) | Map data processing method, device and server | |
GB2542131B (en) | Graphics processing method and system for processing sub-primitives | |
CN110826357A (en) | Method, device, medium and equipment for three-dimensional detection and intelligent driving control of object | |
KR102474160B1 (en) | Map creation method, device, and system, and storage medium | |
CN110853488B (en) | POI (Point of interest) label display method, device and equipment | |
EP2973428B1 (en) | System and method for classification of three-dimensional models in a virtual environment | |
JP5893445B2 (en) | Image processing apparatus and method of operating image processing apparatus | |
US8395824B2 (en) | Method for determining ground line | |
CN105051789A (en) | Three-dimensional map display system | |
CN114648640A (en) | Target object monomer method, device, equipment and storage medium | |
US20220051055A1 (en) | Training data generation method and training data generation device | |
CN112150538B (en) | Method and device for determining vehicle pose in three-dimensional map construction process | |
CN110046623B (en) | Image feature point extraction method and camera | |
CN114241105A (en) | Interface rendering method, device, equipment and computer readable storage medium | |
CN113468363A (en) | Target detection acceleration method for remote sensing image | |
EP3825730A1 (en) | Assigning each point of a point cloud to a scanner position of a plurality of different scanner positions in a point cloud | |
CN1936957B (en) | Optimization and view dependency reduction for processing slice-based volumes | |
CN116481517A (en) | Extended mapping method, device, computer equipment and storage medium | |
JPH08137908A (en) | Method and device for retrieving picture | |
CN112848312B (en) | Method and device for detecting three-dimensional model object, computer equipment and storage medium | |
CN114898321A (en) | Method, device, equipment, medium and system for detecting road travelable area | |
CN115147442A (en) | Grid pattern vectorization method, mobile terminal, electronic device, and medium | |
JPH01150981A (en) | Three-dimensional graphic display device |
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 | ||
CB03 | Change of inventor or designer information | ||
CB03 | Change of inventor or designer information |
Inventor after: Zhang Lan Inventor after: Yang Linzhuo Inventor after: Li Xiangyang Inventor before: Zhang Lan Inventor before: Li Xiangyang Inventor before: Yang Linzhuo |
|
GR01 | Patent grant | ||
GR01 | Patent grant |