CN111951189B - Data enhancement method for multi-scale texture randomization - Google Patents

Data enhancement method for multi-scale texture randomization Download PDF

Info

Publication number
CN111951189B
CN111951189B CN202010813012.1A CN202010813012A CN111951189B CN 111951189 B CN111951189 B CN 111951189B CN 202010813012 A CN202010813012 A CN 202010813012A CN 111951189 B CN111951189 B CN 111951189B
Authority
CN
China
Prior art keywords
img
sample
image
frame
samples
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.)
Active
Application number
CN202010813012.1A
Other languages
Chinese (zh)
Other versions
CN111951189A (en
Inventor
井焜
陈英鹏
许野平
刘辰飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Synthesis Electronic Technology Co Ltd
Original Assignee
Synthesis Electronic Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Synthesis Electronic Technology Co Ltd filed Critical Synthesis Electronic Technology Co Ltd
Priority to CN202010813012.1A priority Critical patent/CN111951189B/en
Publication of CN111951189A publication Critical patent/CN111951189A/en
Application granted granted Critical
Publication of CN111951189B publication Critical patent/CN111951189B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4084—Scaling of whole images or parts thereof, e.g. expanding or contracting in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
    • 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
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G06T7/46—Analysis of texture based on statistical description of texture using random fields
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2200/00—Indexing scheme for image data processing or generation, in general
    • G06T2200/32—Indexing scheme for image data processing or generation, in general involving image mosaicing
    • 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30204—Marker
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a data enhancement method for multi-scale texture randomization, which splices 4 random training samples to form an output sample, and the output sample retains the characteristics of the 4 training samples, thereby increasing the sample characteristics and preventing overfitting during training; and randomly increasing a texture mask frame, comparing the self-checking overlapping area of the texture mask frame and the sample mark frame, if the overlapping area is smaller than a set threshold value, reserving the mask frame, and if the sample mark frame is an overlapping area mark frame, realizing the processing of the overlapping area. According to the invention, through carrying out multi-scale texture randomization data enhancement on the training sample, the preprocessing effect of the training data of the target detection task can be improved, and the recognition detection effect is improved.

Description

Data enhancement method for multi-scale texture randomization
Technical Field
The invention relates to the field of artificial intelligence, in particular to a computer vision typical target detection task training data preprocessing stage, and specifically relates to a data enhancement method for multi-scale texture randomization.
Background
In a real application scene, a large number of occlusion problems often exist. In the training samples of the target detection task, a large number of marked targets are overlapped, so that the shielded targets exist, and the characteristics of part of other targets exist in the training process, thereby influencing the recognition and detection effects of the targets.
A data enhancement method is proposed in the paper Improved reconstruction of volumetric Neural Networks with cut (https:// axiv. org/abs/1708.04552) to crop at random locations and areas of a certain size on an image. The method adds the occluded samples in the training as much as possible, but cannot be used for well managing the condition that a large number of occluded samples exist in the training samples.
A data enhancement method is proposed in a paper mixup, which is a book of radial music task minimizio (https:// axiv. org/abs/1710.09412), and two pictures are randomly selected to be superposed. In the target detection task, the samples with overlapping area overlapping proportion are not effectively distinguished. The method aims to generate more samples in a combined mode, and the condition that a large number of shielding samples exist in training samples cannot be well understood.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a multi-scale texture randomization data enhancement method, which increases sample characteristics, fills a random texture mask frame in a sample region and increases the effect of training data preprocessing of a target detection task.
In order to solve the technical problem, the sampling technical characteristics of the invention are as follows:
a multi-scale texture randomized data enhancement method comprises the following steps:
s01), selecting N training samples P = { P = { (P))0,P1,…,Pn-1And marker box information T = { [ X ] corresponding to N training samples0,Y0,W0,H0,L0],[X1,Y1,W1,H1,L1],…,[Xn-1,Yn-1,Wn-1,Hn-1,Ln-1]}；
Wherein P contains picture information, e.g. P0Contains (img 0, img _ w0, img _ h 0), img0 represents image P0Img _ w0 denotes a picture P0Img _ h0 represents the image P0High, P1、…、Pn-1 The same as above;
t contains mark information of the image, [ X, Y, W, H, L ] is represented as a set of mark frame information, respectively representing the upper left corner (X, Y) of the mark frame, W is width, H is height, and L is the category of the frame;
s02), randomly selecting 4 samples from the training sample set and marking frame information corresponding to the 4 samples, and recording the 4 samples as Ptl、Ptr、Pbl、PbrAnd marking the mark frame information corresponding to 4 samples as Xtl,Ytl,Wtl,Htl,Ltl], [Xtr,Ytr,Wtr,Htr,Ltr], [Xbl,Ybl,Wbl,Hbl,Lbl], [Xbr,Ybr,Wbr,Hbr,Lbr]；
S03), randomly generating 4 scaling factors, i.e., generating S = [ S0, S1, S2, S3], wherein S ranges between [0.5,1.0 ];
s04), and setting the sample after data enhancement as PoutSetting up PoutThe image information of (img, img _ w, img _ h), wherein img represents the data-enhanced image PoutImg _ w represents the picture PoutImg _ h represents the picture PoutThe height of (d);
s05), and setting the coordinates of the center point of the sample Pout after data enhancement as (xc, yc), then
xc = img_w / 2 + b （1），
yc = img_h / 2 + b （2），
Wherein b [ - (img _ w + img _ h)/16, (img _ w + img _ h)/16 ];
s06), when the input sample is PtlTime, image imgtlMultiplying by a scaling factor s0, changing the image scale, and marking the output image as Ptl0And similarly, obtaining an output image through scaling of other three input samples and recording the output image as Ptr1、Pbl2、Pbr3Output sample PoutIs Ptl0、Ptr1、Pbl2、Pbr3Splicing in different scales, and converting corresponding mark frame information to be recorded as Tout；
S07), randomly generating n texture Mask boxes with different sizes, different shapes and different colors, and recording as Mask = [ m ]0,m1,…,mn-1]；
S08), calculating an Overlap area of each generated mask frame and output sample mark frame, the Overlap area being denoted as Overlap = [ o =0,o1,…,on-1]；
S09), hypothesis randomGenerating a mask frame miAnd a certain mark frame t in the output samplejOverlap, the area of the overlap region is areaiThen oi = areai / （wj * hj) Wherein the frame t is markedjIs [ x ] as position informationj,yj,wj,hj];
S10), value o when overlapping areaiGreater than a threshold value ostThen the mask frame m is deletedi。
Further, Ptl0、Ptr1、Pbl2、Pbr3Splicing different scales to form an output sample PoutThe process comprises the following steps:
s61), placing the Ptl0 sample partial area at the upper left corner of Pout, and the concrete conversion formula is:
x1a = max(xc - imgtl0_w, 0)，
y1a = max(yc - imgtl0_h, 0)，
x2a = xc，
y2a = yc，
x1b = imgtl0_w – (x2a – x1a)，
y1b = imgtl0_h – (y2a – y1a)，
x2b = imgtl0_w，
y2b = imgtl0_h，
img[y1a:y2a, x1a:x2a] = imgtl0[y1b:y2b, x1b:x2b]；
s62), adding Ptr1Partial area placement of sample PoutThe specific conversion formula of the upper right corner of the table is as follows:
x1a = xc，
y1a = max(yc - imgtr1_h, 0)，
x2a = min(xc + imgtr1_w, img_w)，
y2a = yc，
x1b = 0，
y1b = imgtr1_h – (y2a – y1a)，
x2b = min(imgtr1_w, x2a –x1a)，
y2b = imgtr1_h，
img[y1a:y2a, x1a:x2a] = imgtr1[y1b:y2b, x1b:x2b]；
s63), adding Pbl2Partial area placement P of sampleoutThe specific conversion formula is:
x1a = max(xc – imgbl2_w, 0)，
y1a = yc，
x2a = xc，
y2a = min(img_h, yc + imgbl2_h)，
x1b = imgbl2_w – (x2a – x1a)，
y1b = 0，
x2b = max(xc, imgbl2_w)，
y2b = min(y2a – y1a ,imgbl2_h)，
img[y1a:y2a, x1a:x2a] = imgbl2[y1b:y2b, x1b:x2b]；
s64), adding Pbr3Partial area placement P of sampleoutThe specific conversion formula is as follows:
x1a = xc，
y1a = yc，
x2a = min(xc + imgbr3_w, img_w)，
y2a = min(img_h, yc + imgbr3_h)，
x1b = 0，
y1b = 0，
x2b = min(imgbr3_w, x2a –x1a)，
y2b = min(y2a – y1a ,imgbr3_h)，
img[y1a:y2a, x1a:x2a] = imgbr3[y1b:y2b, x1b:x2b]。
further, ostThe value of (d) was chosen to be 0.5.
The invention has the beneficial effects that: according to the invention, 4 random training samples are spliced to form an output sample, the output sample retains the characteristics of the 4 training samples, the sample characteristics can be increased, and overfitting during training is prevented; and randomly increasing a texture mask frame, comparing the self-checking overlapping area of the texture mask frame and the sample mark frame, if the overlapping area is smaller than a set threshold value, reserving the mask frame, and if the sample mark frame is an overlapping area mark frame, realizing the processing of the overlapping area. According to the invention, through carrying out multi-scale texture randomization data enhancement on the training sample, the preprocessing effect of the training data of the target detection task can be improved, and the recognition detection effect is improved.
Drawings
Fig. 1 is a schematic diagram of output samples formed by splicing 4 training samples in example 1.
Detailed Description
The invention is further described with reference to the following figures and specific examples.
Example 1
The embodiment discloses a data enhancement method for multi-scale texture randomization, which comprises the following steps:
s01), selecting N training samples P = { P = { (P))0,P1,…,Pn-1And marker box information T = { [ X ] corresponding to N training samples0,Y0,W0,H0,L0],[X1,Y1,W1,H1,L1],…,[Xn-1,Yn-1,Wn-1,Hn-1,Ln-1]}；
Wherein P contains picture information, e.g. P0Contains (img 0, img _ w0, img _ h 0), img0 represents image P0Img _ w0 denotes a picture P0Img _ h0 represents the image P0High, P1、…、Pn-1 Has the same meaning as above;
t contains mark information of an image, as shown in FIG. 1, a plurality of mark frames exist in one training sample, namely one picture, so [ X, Y, W, H, L ] represents a group of mark frame information and respectively represents the upper left corner (X, Y) of the mark frame, W is width, H is height, and L is the category of the frame;
s02), randomly selecting 4 samples from the training sample set and marking frame information corresponding to the 4 samples, and recording the 4 samples as Ptl、Ptr、Pbl、PbrMarks corresponding to 4 samplesThe frame information is marked as [ X ]tl,Ytl,Wtl,Htl,Ltl], [Xtr,Ytr,Wtr,Htr,Ltr], [Xbl,Ybl,Wbl,Hbl,Lbl], [Xbr,Ybr,Wbr,Hbr,Lbr]；
S03), randomly generating 4 scale factors, namely generating S = [ S0, S1, S2, S3], wherein the range of S is between [0.5 and 1.0], namely S ∈ [0.5 and 1.0 ];
s04), and setting the sample after data enhancement as PoutSetting up PoutThe image information of (img, img _ w, img _ h), wherein img represents the data-enhanced image PoutImg _ w represents the picture PoutImg _ h represents the picture PoutThe height of (d);
s05), and setting the coordinates of the center point of the sample Pout after data enhancement as (xc, yc), then
xc = img_w / 2 + b （1），
yc = img_h / 2 + b （2），
Wherein b [ - (img _ w + img _ h)/16, (img _ w + img _ h)/16 ];
s06), when the input sample is PtlTime, image imgtlMultiplying by a scaling factor s0 to change the image scale, and recording the output image as Ptl0Similarly, the other three input samples are scaled to obtain an output image which is recorded as Ptr1、Pbl2、Pbr3Output sample PoutIs Ptl0、Ptr1、Pbl2、Pbr3Splicing in different scales, and converting corresponding mark frame information to be recorded as Tout；
S07), randomly generating n texture Mask frames with different sizes, different shapes and different colors, and recording the frames as Mask = [ m ]0,m1,…,mn-1]；
S08), calculating an Overlap area of each generated mask frame and output sample mark frame, the Overlap area being denoted as Overlap = [ o =0,o1,…,on-1]；
S09), assuming random generation of mask frame miAnd a certain mark frame t in the output samplejOverlap, the area of the overlap region is areaiThen o isi = areai / （wj * hj) Wherein the frame t is markedjIs [ x ] as position informationj,yj,wj,hj];
S10), value o when overlapping areaiGreater than a threshold value ostThen the mask frame m is deletediGeneral ostThe value of (d) was chosen to be 0.5.
As shown in FIG. 1, Ptl0、Ptr1、Pbl2、Pbr3Splicing different scales to form an output sample PoutThe process comprises the following steps:
s61), placing the part area of Ptl0 samples at the upper left corner of Pout, and the concrete conversion formula is:
x1a = max(xc - imgtl0_w, 0)，
y1a = max(yc - imgtl0_h, 0)，
x2a = xc，
y2a = yc，
x1b = imgtl0_w – (x2a – x1a)，
y1b = imgtl0_h – (y2a – y1a)，
x2b = imgtl0_w，
y2b = imgtl0_h，
img[y1a:y2a, x1a:x2a] = imgtl0[y1b:y2b, x1b:x2b]；
wherein, (x1a, y1a), (x2a, y2a) are images PoutThe coordinates of the upper left corner and the lower right corner of part A1, (x1b, y1b), (x2b, y2b) are respectively an image Ptl0Coordinates of the upper left corner and the lower right corner of part a2, and the last formula shows that image P is to be processedtl0Is mapped to image PoutPart a 1.
S62), adding Ptr1Partial area placement of sample PoutThe specific conversion formula of the upper right corner of the table is as follows:
x1a = xc，
y1a = max(yc - imgtr1_h, 0)，
x2a = min(xc + imgtr1_w, img_w)，
y2a = yc，
x1b = 0，
y1b = imgtr1_h – (y2a – y1a)，
x2b = min(imgtr1_w, x2a –x1a)，
y2b = imgtr1_h，
img[y1a:y2a, x1a:x2a] = imgtr1[y1b:y2b, x1b:x2b]；
wherein, (x1a, y1a), (x2a, y2a) are images PoutThe coordinates of the upper left corner and the lower right corner of part B1, (x1B, y1B), (x2B, y2B) are respectively an image Ptr1B2 part, the last formula representing image Ptr1Is mapped to image PoutPart B1.
S63), adding Pbl2Partial area placement of sample PoutThe specific conversion formula is:
x1a = max(xc – imgbl2_w, 0)，
y1a = yc，
x2a = xc，
y2a = min(img_h, yc + imgbl2_h)，
x1b = imgbl2_w – (x2a – x1a)，
y1b = 0，
x2b = max(xc, imgbl2_w)，
y2b = min(y2a – y1a ,imgbl2_h)，
img[y1a:y2a, x1a:x2a] = imgbl2[y1b:y2b, x1b:x2b]；
wherein, (x1a, y1a), (x2a, y2a) are images PoutThe coordinates of the upper left corner and the lower right corner of the C1 part of (x1b, y1b), (x2b, y2b) are respectively the image Pbl2C2, and the last formula represents image Pbl2Is mapped to image PoutPart C1.
S64), adding Pbr3Partial area placement of sample PoutThe specific conversion formula is as follows:
x1a = xc，
y1a = yc，
x2a = min(xc + imgbr3_w, img_w)，
y2a = min(img_h, yc + imgbr3_h)，
x1b = 0，
y1b = 0，
x2b = min(imgbr3_w, x2a –x1a)，
y2b = min(y2a – y1a ,imgbr3_h)，
img[y1a:y2a, x1a:x2a] = imgbr3[y1b:y2b, x1b:x2b];
wherein, (x1a, y1a), (x2a, y2a) are images PoutThe coordinates of the upper left corner and the lower right corner of the D1 part of (x1b, y1b), (x2b, y2b) are the image Pbr3The coordinates of the upper left and lower right parts of part D2, and the last formula represents image Pbr3Is mapped to image PoutPart D1.
According to the invention, 4 random training samples are spliced to form an output sample, the output sample retains the characteristics of the 4 training samples, the sample characteristics can be increased, and overfitting during training is prevented; and randomly increasing a texture mask frame, comparing the self-checking overlapping area of the texture mask frame and the sample mark frame, if the overlapping area is smaller than a set threshold value, reserving the mask frame, and if the sample mark frame is the overlapping area mark frame, realizing the processing of the overlapping area. According to the invention, through carrying out multi-scale texture randomization data enhancement on the training samples, the effect of training data preprocessing of the target detection task can be improved, and the recognition detection effect is improved.
The foregoing description is only for the purpose of illustrating the general principles and preferred embodiments of the present invention, and it is intended that modifications and substitutions be made by those skilled in the art in light of the present invention and that they fall within the scope of the present invention.

Claims (3)

1. A data enhancement method of multi-scale texture randomization is characterized in that: the method comprises the following steps:
s01), selecting N training samples P = { P = { (P))0,P1,…,Pn-1And marker box information T = { [ X ] corresponding to N training samples0,Y0,W0,H0,L0],[X1,Y1,W1,H1,L1],…,[Xn-1,Yn-1,Wn-1,Hn-1,Ln-1]}；
Wherein P contains picture information, e.g. P0Contains (img 0, img _ w0, img _ h 0), img0 represents image P0Img _ w0 denotes a picture P0Img _ h0 represents the image P0High, P1、…、Pn-1 The same as above;
t contains mark information of the image, [ X, Y, W, H, L ] is represented as a set of mark frame information, respectively representing the upper left corner (X, Y) of the mark frame, W is width, H is height, and L is the category of the frame;
s02), randomly selecting 4 samples from the training sample set and marking frame information corresponding to the 4 samples, and recording the 4 samples as Ptl、Ptr、Pbl、PbrAnd marking the mark frame information corresponding to 4 samples as Xtl,Ytl,Wtl,Htl,Ltl], [Xtr,Ytr,Wtr,Htr,Ltr], [Xbl,Ybl,Wbl,Hbl,Lbl], [Xbr,Ybr,Wbr,Hbr,Lbr]；
S03), randomly generating 4 scaling factors, i.e., generating S = [ S0, S1, S2, S3], wherein S ranges between [0.5,1.0 ];
s04), and setting the sample after data enhancement as PoutSetting up PoutThe image information of (img, img _ w, img _ h), wherein img represents the data-enhanced image PoutImg _ w represents the picture PoutImg _ h represents the picture PoutThe height of (d);
s05), and setting the data enhanced sample PoutThe coordinate of the center point is (xc, yc)Then, then
xc = img_w / 2 + b （1），
yc = img_h / 2 + b （2），
Wherein b [ - (img _ w + img _ h)/16, (img _ w + img _ h)/16 ];
s06), when the input sample is PtlTime, image imgtlMultiplying by a scaling factor s0 to change the image scale, and recording the output image as Ptl0And similarly, obtaining an output image through scaling of other three input samples and recording the output image as Ptr1、Pbl2、Pbr3Output sample PoutIs Ptl0、Ptr1、Pbl2、Pbr3Splicing in different scales, and converting corresponding mark frame information to be recorded as Tout；
S07), randomly generating n texture Mask frames with different sizes, different shapes and different colors, and recording the frames as Mask = [ m ]0,m1,…,mn-1]；
S08), calculating an Overlap area of each generated mask frame and output sample mark frame, the Overlap area being denoted as Overlap = [ o =0,o1,…,on-1]；
S09), assuming random generation of mask frame miAnd a certain mark frame t in the output samplejOverlap, the area of the overlap region is areaiThen oi = areai / （wj * hj) Wherein the frame t is markedj Is [ x ] as position informationj,yj,wj,hj];
S10), value o when overlapping areaiGreater than a threshold value ost Then the mask frame m is deletedi。
2. The method of multi-scale texture randomization of data enhancement as recited in claim 1, wherein: ptl0、Ptr1、Pbl2、Pbr3Splicing different scales to form an output sample PoutThe process comprises the following steps:
s61), willPtl0Partial area placement of sample PoutThe specific conversion formula of the upper left corner of (1) is as follows:
x1a = max(xc - imgtl0_w, 0)，
y1a = max(yc - imgtl0_h, 0)，
x2a = xc，
y2a = yc，
x1b = imgtl0_w – (x2a – x1a)，
y1b = imgtl0_h – (y2a – y1a)，
x2b = imgtl0_w，
y2b = imgtl0_h，
img[y1a:y2a, x1a:x2a] = imgtl0[y1b:y2b, x1b:x2b]；
s62), adding Ptr1Partial area placement of sample PoutThe specific conversion formula of the upper right corner of the table is as follows:
x1a = xc，
y1a = max(yc - imgtr1_h, 0)，
x2a = min(xc + imgtr1_w, img_w)，
y2a = yc，
x1b = 0，
y1b = imgtr1_h – (y2a – y1a)，
x2b = min(imgtr1_w, x2a –x1a)，
y2b = imgtr1_h，
img[y1a:y2a, x1a:x2a] = imgtr1[y1b:y2b, x1b:x2b]；
s063), mixing Pbl2Partial area placement of sample PoutThe specific conversion formula is:
x1a = max(xc – imgbl2_w, 0)，
y1a = yc，
x2a = xc，
y2a = min(img_h, yc + imgbl2_h)，
x1b = imgbl2_w – (x2a – x1a)，
y1b = 0，
x2b = max(xc, imgbl2_w)，
y2b = min(y2a – y1a ,imgbl2_h)，
img[y1a:y2a, x1a:x2a] = imgbl2[y1b:y2b, x1b:x2b]；
s064), adding Pbr3Partial area placement of sample PoutThe concrete conversion formula of the lower right corner of the table is as follows:
x1a = xc，
y1a = yc，
x2a = min(xc + imgbr3_w, img_w)，
y2a = min(img_h, yc + imgbr3_h)，
x1b = 0，
y1b = 0，
x2b = min(imgbr3_w, x2a –x1a)，
y2b = min(y2a – y1a ,imgbr3_h)，
img[y1a:y2a, x1a:x2a] = imgbr3[y1b:y2b, x1b:x2b]。
3. the method of multi-scale texture randomization of data enhancement of claim 1, characterized in that: ostThe value of (d) was chosen to be 0.5.
CN202010813012.1A 2020-08-13 2020-08-13 Data enhancement method for multi-scale texture randomization Active CN111951189B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010813012.1A CN111951189B (en) 2020-08-13 2020-08-13 Data enhancement method for multi-scale texture randomization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010813012.1A CN111951189B (en) 2020-08-13 2020-08-13 Data enhancement method for multi-scale texture randomization

Publications (2)

Publication Number Publication Date
CN111951189A CN111951189A (en) 2020-11-17
CN111951189B true CN111951189B (en) 2022-05-06

Family

ID=73341855

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010813012.1A Active CN111951189B (en) 2020-08-13 2020-08-13 Data enhancement method for multi-scale texture randomization

Country Status (1)

Country Link
CN (1) CN111951189B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112348744B (en) * 2020-11-24 2022-07-01 电子科技大学 Data enhancement method based on thumbnail
CN112967187B (en) * 2021-02-25 2024-05-31 深圳海翼智新科技有限公司 Method and apparatus for target detection

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107085609A (en) * 2017-04-24 2017-08-22 国网湖北省电力公司荆州供电公司 A kind of pedestrian retrieval method that multiple features fusion is carried out based on neutral net
CN108921817A (en) * 2018-05-24 2018-11-30 浙江工业大学 A kind of data enhancement methods for skin disease image
GB201918431D0 (en) * 2019-12-13 2020-01-29 British Broadcasting Corp Video encoding and video decoding
CN111179253A (en) * 2019-12-30 2020-05-19 歌尔股份有限公司 Product defect detection method, device and system

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107085609A (en) * 2017-04-24 2017-08-22 国网湖北省电力公司荆州供电公司 A kind of pedestrian retrieval method that multiple features fusion is carried out based on neutral net
CN108921817A (en) * 2018-05-24 2018-11-30 浙江工业大学 A kind of data enhancement methods for skin disease image
GB201918431D0 (en) * 2019-12-13 2020-01-29 British Broadcasting Corp Video encoding and video decoding
CN111179253A (en) * 2019-12-30 2020-05-19 歌尔股份有限公司 Product defect detection method, device and system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
FMix: Enhancing Mixed Sample Data Augmentation;Ethan Harris 等;《https://arxiv.org/abs/2002.12047》;20200227;第1-9页 *

Also Published As

Publication number Publication date
CN111951189A (en) 2020-11-17

Similar Documents

Publication Publication Date Title
US7254270B2 (en) System and method for bounding and classifying regions within a graphical image
CN111951189B (en) Data enhancement method for multi-scale texture randomization
CN109284757A (en) A kind of licence plate recognition method, device, computer installation and computer readable storage medium
CN109784227B (en) image detection and identification method and device
CN112541922A (en) Test paper layout segmentation method based on digital image, electronic equipment and storage medium
CN112699885A (en) Semantic segmentation training data augmentation method and system based on antagonism generation network GAN
CN110569379A (en) Method for manufacturing picture data set of automobile parts
CN111833362A (en) Unstructured road segmentation method and system based on superpixel and region growing
Huang et al. Obmo: One bounding box multiple objects for monocular 3d object detection
CN115861733A (en) Point cloud data labeling method, model training method, electronic device and storage medium
EP3591620A1 (en) Image processing device and two-dimensional image generation program
CN115270184A (en) Video desensitization method, vehicle video desensitization method and vehicle-mounted processing system
CN117830537B (en) Weak supervision 3D scene graph generation method, device, equipment and medium
CN113808142A (en) Ground identifier identification method and device and electronic equipment
CN108230227A (en) A kind of recognition methods of distorted image, device and electronic equipment
CN112434581A (en) Outdoor target color identification method and system, electronic device and storage medium
CN116168393B (en) Automatic semantic annotation data generation method and device based on point cloud neural radiation field
CN116091784A (en) Target tracking method, device and storage medium
AU2021240228A1 (en) Method, apparatus and device for recognizing stacked objects, and computer storage medium
CN114066933A (en) Multi-target tracking method, system and related equipment
TWI807904B (en) Method for training depth identification model, method for identifying depth of images and related devices
WO2024197829A1 (en) Single-character detection method and apparatus, model training method and apparatus, and device and medium
CN115082409B (en) Automatic change system of discernment nuclide image diagnosis myocardial ischemia
CN111047604B (en) Transparency mask extraction method and device for high-definition image and storage medium
TWI715184B (en) Method and device for positioning target object in image, computer device and storage medium

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
GR01 Patent grant
GR01 Patent grant