Caltech Pedestrian Detection Benchmark
阿新 • • 發佈：2018-12-09
環境：ubuntu，matlab
下載toolbox及評估程式碼
#!/bin/bash # download toolbox git clone https://github.com/pdollar/toolbox pdollar_toolbox # download evaluation/labeling code wget http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/code/code3.2.1.zip mkdir code unzip code3.2.1.zip -d code rm code3.2.1.zip #download annotations cd code mkdir data-USA cd data-USA wget http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/datasets/USA/annotations.zip unzip annotations.zip rm annotations.zip #download results mkdir res cd res list=("VJ" "HOG" "FPDW" "RPN+BF" "MS-CNN" "SDS-RCNN" "TLL-TFA" "F-DNN2+SS" "F-DNN+SS" "GDFL" "PCN" "F-DNN" "AdaptFasterRCNN" "SA-FastRCNN" "UDN+" "FasterRCNN+ATT") for i in "${list[@]}" do http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/datasets/USA/res/${method}.zip unzip ${method}.zip rm ${method}.zip done
評估自己的實驗結果
一、通用步驟
1、依照官網要求的格式處理自己的資料，放到code/data-USA/res資料夾下。
官網要求的格式：res/目錄下新建一個資料夾命名為自己方法的名字（如：my_method），res/my_method/set06/V000/I00029.txt，每個txt檔案記憶體放對應圖片的檢測結果，每行的格式為[left, top, width, height]
2、用matlab評估
# 第一次使用 p = genpath('../pdollar_toolbox'); addpath(p); #畫MR-FPPI圖 將演算法結果放入./data-USA/res/XXX/ 修改dbEval.m檔案，exps只保留Reasonble，變數algs只保留你需要的方法、然後加上自己的方法，變數dataNames只保留UsaTest 最終生成的結果會存放到code/results資料夾下
二、YOLOv3檢測行人的結果
exps = Reasonble時，MR=11%，在所有的方法裡面排在第13名。
這個結果是博主自己用yolov3只檢測行人，然後在評估程式碼上跑出來的。
將下載後的檔案解壓放到data-USA/res/資料夾下，然後如下修改dbEval.m檔案，最後執行dbEval.m
function dbEval
% Evaluate and plot all pedestrian detection results.
%
% Set parameters by altering this function directly.
%
% USAGE
% dbEval
%
% INPUTS
%
% OUTPUTS
%
% EXAMPLE
% dbEval
%
% See also bbGt, dbInfo
%
% Caltech Pedestrian Dataset Version 3.2.1
% Copyright 2014 Piotr Dollar. [pdollar-at-gmail.com]
% Licensed under the Simplified BSD License [see external/bsd.txt]
% List of experiment settings: { name, hr, vr, ar, overlap, filter }
% name - experiment name
% hr - height range to test
% vr - visibility range to test
% ar - aspect ratio range to test
% overlap - overlap threshold for evaluation
% filter - expanded filtering (see 3.3 in PAMI11)
exps = {
'Reasonable', [50 inf], [.65 inf], 0, .5, 1.25
'All', [20 inf], [.2 inf], 0, .5, 1.25
% 'Scale=large', [100 inf], [inf inf], 0, .5, 1.25
'Scale=near', [80 inf], [inf inf], 0, .5, 1.25
'Scale=medium', [30 80], [inf inf], 0, .5, 1.25
'Scale=far', [20 30], [inf inf], 0, .5, 1.25
'Occ=none', [50 inf], [inf inf], 0, .5, 1.25
'Occ=partial', [50 inf], [.65 1], 0, .5, 1.25
'Occ=heavy', [50 inf], [.2 .65], 0, .5, 1.25
% 'Ar=all', [50 inf], [inf inf], 0, .5, 1.25
'Ar=typical', [50 inf], [inf inf], .1, .5, 1.25
'Ar=atypical', [50 inf], [inf inf], -.1, .5, 1.25
% 'Overlap=25', [50 inf], [.65 inf], 0, .25, 1.25
% 'Overlap=50', [50 inf], [.65 inf], 0, .50, 1.25
% 'Overlap=75', [50 inf], [.65 inf], 0, .75, 1.25
% 'Expand=100', [50 inf], [.65 inf], 0, .5, 1.00
% 'Expand=125', [50 inf], [.65 inf], 0, .5, 1.25
% 'Expand=150', [50 inf], [.65 inf], 0, .5, 1.50
};
exps=cell2struct(exps',{'name','hr','vr','ar','overlap','filter'});
% List of algorithms: { name, resize, color, style }
% name - algorithm name (defines data location)
% resize - if true rescale height of each box by 100/128
% color - algorithm plot color
% style - algorithm plot linestyle
n=1000; clrs=zeros(n,3);
for i=1:n, clrs(i,:)=max(.3,mod([78 121 42]*(i+1),255)/255); end
algs = {
'VJ', 0, clrs(1,:), '-'
'HOG', 1, clrs(2,:), '--'
% 'FtrMine', 1, clrs(3,:), '-'
% 'Shapelet', 0, clrs(4,:), '--'
% 'PoseInv', 1, clrs(5,:), '-'
% 'MultiFtr', 0, clrs(6,:), '--'
% 'MultiFtr+CSS', 0, clrs(7,:), '-'
% 'MultiFtr+Motion', 0, clrs(8,:), '--'
% 'HikSvm', 1, clrs(9,:), '-'
% 'Pls', 0, clrs(10,:), '--'
% 'HogLbp', 0, clrs(11,:), '-'
% 'LatSvm-V1', 0, clrs(12,:), '--'
% 'LatSvm-V2', 0, clrs(13,:), '-'
% 'ChnFtrs', 0, clrs(14,:), '--'
'FPDW', 0, clrs(15,:), '-'
% 'FeatSynth', 0, clrs(16,:), '--'
% 'MultiResC', 0, clrs(17,:), '-'
% 'CrossTalk', 0, clrs(18,:), '--'
% 'VeryFast', 0, clrs(19,:), '-'
% 'ConvNet', 0, clrs(20,:), '--'
% 'SketchTokens', 0, clrs(21,:), '-'
% 'Roerei', 0, clrs(22,:), '--'
% 'AFS', 1, clrs(23,:), '-'
% 'AFS+Geo', 1, clrs(23,:), '--'
% 'MLS', 1, clrs(24,:), '-'
% 'MT-DPM', 0, clrs(25,:), '-'
% 'MT-DPM+Context', 0, clrs(25,:), '--'
% 'DBN-Isol', 0, clrs(26,:), '-'
% 'DBN-Mut', 0, clrs(26,:), '--'
% 'MF+Motion+2Ped', 0, clrs(27,:), '-'
% 'MultiResC+2Ped', 0, clrs(27,:), '--'
% 'MOCO', 0, clrs(28,:), '-'
% 'ACF', 0, clrs(29,:), '-'
% 'ACF-Caltech', 0, clrs(29,:), '--'
% 'ACF+SDt', 0, clrs(30,:), '-'
% 'FisherBoost', 0, clrs(31,:), '--'
% 'pAUCBoost', 0, clrs(32,:), '-'
% 'Franken', 0, clrs(33,:), '--'
% 'JointDeep', 0, clrs(34,:), '-'
% 'MultiSDP', 0, clrs(35,:), '--'
% 'SDN', 0, clrs(36,:), '-'
% 'RandForest', 0, clrs(37,:), '--'
% 'WordChannels', 0, clrs(38,:), '-'
% 'InformedHaar', 0, clrs(39,:), '--'
% 'SpatialPooling', 0, clrs(40,:), '-'
% 'SpatialPooling+', 0, clrs(42,:), '--'
% 'LDCF', 0, clrs(43,:), '-'
% 'ACF-Caltech+', 0, clrs(44,:), '--'
% 'Katamari', 0, clrs(45,:), '-'
% 'NAMC', 0, clrs(46,:), '--'
% 'FastCF', 0, clrs(47,:), '-'
% 'TA-CNN', 0, clrs(48,:), '--'
% 'SCCPriors', 0, clrs(49,:), '-'
% 'DeepParts', 0, clrs(50,:), '--'
% 'DeepCascade', 0, clrs(51,:), '-'
% 'DeepCascade+', 0, clrs(51,:), '--'
% 'LFOV', 0, clrs(52,:), '-'
% 'Checkerboards', 0, clrs(53,:), '--'
% 'Checkerboards+', 0, clrs(53,:), '-'
% 'CCF', 0, clrs(54,:), '--'
% 'CCF+CF', 0, clrs(54,:), '-'
% 'CompACT-Deep', 0, clrs(55,:), '--'
% 'SCF+AlexNet', 0, clrs(56,:), '-'
'SA-FastRCNN', 0, clrs(57,:), '--'
'RPN+BF', 0, clrs(58,:), '-'
'MS-CNN', 0, clrs(59,:), '--'
% 'ACF++', 0, clrs(60,:), '-'
% 'LDCF++', 0, clrs(61,:), '--'
% 'MRFC+Semantic', 0, clrs(63,:), '--'
'F-DNN', 0, clrs(64,:), '-'
'F-DNN+SS', 0, clrs(65,:), '--'
'UDN+', 0, clrs(66,:), '-'
'SDS-RCNN', 0, clrs(67,:), '--'
'PCN', 0, clrs(68,:), '-'
'F-DNN2+SS', 0, clrs(69,:), '--'
'AdaptFasterRCNN', 0, clrs(70,:), '-'
'FasterRCNN+ATT', 0, clrs(71,:), '--'
'TLL-TFA', 0, clrs(72,:), '-'
'YOLO', 0, clrs(73,:), '--'
};
algs=cell2struct(algs',{'name','resize','color','style'});
% List of database names
% dataNames = {'UsaTest','UsaTrain','InriaTest',...
% 'TudBrussels','ETH','Daimler','Japan'};
dataNames = {'UsaTest'};
% select databases, experiments and algorithms for evaluation
dataNames = dataNames(1); % select one or more databases for evaluation
exps = exps(:); % select one or more experiment for evaluation
algs = algs(:); % select one or more algorithms for evaluation
% remaining parameters and constants
aspectRatio = .41; % default aspect ratio for all bbs
bnds = [5 5 635 475]; % discard bbs outside this pixel range
plotRoc = 1; % if true plot ROC else PR curves
plotAlg = 0; % if true one plot per alg else one plot per exp
plotNum = 15; % only show best plotNum curves (and VJ and HOG)
samples = 10.^(-2:.25:0); % samples for computing area under the curve
lims = [2e-4 50 .035 1]; % axis limits for ROC plots
bbsShow = 0; % if true displays sample bbs for each alg/exp
bbsType = 'fp'; % type of bbs to display (fp/tp/fn/dt)
algs0=algs; bnds0=bnds;
for d=1:length(dataNames), dataName=dataNames{d};
% select algorithms with results for current dataset
[~,set]=dbInfo(dataName); set=['/set' int2str2(set(1),2)];
names={algs0.name}; n=length(names); keep=false(1,n);
for i=1:n, keep(i)=exist([dbInfo '/res/' names{i} set],'dir'); end
algs=algs0(keep);
% handle special database specific cases
if(any(strcmp(dataName,{'InriaTest','TudBrussels','ETH'})))
bnds=[-inf -inf inf inf]; else bnds=bnds0; end
if(strcmp(dataName,'InriaTest'))
i=find(strcmp({algs.name},'FeatSynth'));
if(~isempty(i)), algs(i).resize=1; end;
end
% name for all plots (and also temp directory for results)
plotName=[fileparts(mfilename('fullpath')) '/results/' dataName];
if(~exist(plotName,'dir')), mkdir(plotName); end
% load detections and ground truth and evaluate
dts = loadDt( algs, plotName, aspectRatio );
gts = loadGt( exps, plotName, aspectRatio, bnds );
res = evalAlgs( plotName, algs, exps, gts, dts );
% plot curves and bbs
plotExps( res, plotRoc, plotAlg, plotNum, plotName, ...
samples, lims, reshape([algs.color]',3,[])', {algs.style} );
plotBbs( res, plotName, bbsShow, bbsType );
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function res = evalAlgs( plotName, algs, exps, gts, dts )
% Evaluate every algorithm on each experiment
%
% OUTPUTS
% res - nGt x nDt cell of all evaluations, each with fields
% .stra - string identifying algorithm
% .stre - string identifying experiment
% .gtr - [n x 1] gt result bbs for each frame [x y w h match]
% .dtr - [n x 1] dt result bbs for each frame [x y w h score match]
fprintf('Evaluating: %s\n',plotName); nGt=length(gts); nDt=length(dts);
res=repmat(struct('stra',[],'stre',[],'gtr',[],'dtr',[]),nGt,nDt);
for g=1:nGt
for d=1:nDt
gt=gts{g}; dt=dts{d}; n=length(gt); assert(length(dt)==n);
stra=algs(d).name; stre=exps(g).name;
fName = [plotName '/ev-' [stre '-' stra] '.mat'];
if(exist(fName,'file')), R=load(fName); res(g,d)=R.R; continue; end
fprintf('\tExp %i/%i, Alg %i/%i: %s/%s\n',g,nGt,d,nDt,stre,stra);
hr = exps(g).hr.*[1/exps(g).filter exps(g).filter];
for f=1:n, bb=dt{f}; dt{f}=bb(bb(:,4)>=hr(1) & bb(:,4)<hr(2),:); end
[gtr,dtr] = bbGt('evalRes',gt,dt,exps(g).overlap);
R=struct('stra',stra,'stre',stre,'gtr',{gtr},'dtr',{dtr});
res(g,d)=R; save(fName,'R');
end
end
end
function plotExps( res, plotRoc, plotAlg, plotNum, plotName, ...
samples, lims, colors, styles )
% Plot all ROC or PR curves.
%
% INPUTS
% res - output of evalAlgs
% plotRoc - if true plot ROC else PR curves
% plotAlg - if true one plot per alg else one plot per exp
% plotNum - only show best plotNum curves (and VJ and HOG)
% plotName - filename for saving plots
% samples - samples for computing area under the curve
% lims - axis limits for ROC plots
% colors - algorithm plot colors
% styles - algorithm plot linestyles
% Compute (xs,ys) and score (area under the curve) for every exp/alg
[nGt,nDt]=size(res); xs=cell(nGt,nDt); ys=xs; scores=zeros(nGt,nDt);
for g=1:nGt
for d=1:nDt
[xs{g,d},ys{g,d},~,score] = ...
bbGt('compRoc',res(g,d).gtr,res(g,d).dtr,plotRoc,samples);
if(plotRoc), ys{g,d}=1-ys{g,d}; score=1-score; end
if(plotRoc), score=exp(mean(log(score))); else score=mean(score); end
scores(g,d)=score;
end
end
% Generate plots
if( plotRoc ), fName=[plotName 'Roc']; else fName=[plotName 'Pr']; end
stra={res(1,:).stra}; stre={res(:,1).stre}; scores1=round(scores*100);
if( plotAlg ), nPlots=nDt; else nPlots=nGt; end; plotNum=min(plotNum,nDt);
for p=1:nPlots
% prepare xs1,ys1,lgd1,colors1,styles1,fName1 according to plot type
if( plotAlg )
xs1=xs(:,p); ys1=ys(:,p); fName1=[fName stra{p}]; lgd1=stre;
for g=1:nGt, lgd1{g}=sprintf('%2i%% %s',scores1(g,p),stre{g}); end
colors1=uniqueColors(1,max(10,nGt)); styles1=repmat({'-','--'},1,nGt);
else
xs1=xs(p,:); ys1=ys(p,:); fName1=[fName stre{p}]; lgd1=stra;
for d=1:nDt, lgd1{d}=sprintf('%2i%% %s',scores1(p,d),stra{d}); end
kp=[find(strcmp(stra,'VJ')) find(strcmp(stra,'HOG')) 1 1];
[~,ord]=sort(scores(p,:)); kp=ord==kp(1)|ord==kp(2);
j=find(cumsum(~kp)>=plotNum-2); kp(1:j(1))=1; ord=fliplr(ord(kp));
xs1=xs1(ord); ys1=ys1(ord); lgd1=lgd1(ord); colors1=colors(ord,:);
styles1=styles(ord); f=fopen([fName1 '.txt'],'w');
for d=1:nDt, fprintf(f,'%s %f\n',stra{d},scores(p,d)); end; fclose(f);
end
% plot curves and finalize display
figure(1); clf; grid on; hold on; n=length(xs1); h=zeros(1,n);
for i=1:n, h(i)=plot(xs1{i},ys1{i},'Color',colors1(i,:),...
'LineStyle',styles1{i},'LineWidth',2); end
if( plotRoc )
yt=[.05 .1:.1:.5 .64 .8]; ytStr=int2str2(yt*100,2);
for i=1:length(yt), ytStr{i}=['.' ytStr{i}]; end
set(gca,'XScale','log','YScale','log',...
'YTick',[yt 1],'YTickLabel',[ytStr '1'],...
'XMinorGrid','off','XMinorTic','off',...
'YMinorGrid','off','YMinorTic','off');
xlabel('false positives per image','FontSize',14);
ylabel('miss rate','FontSize',14); axis(lims);
else
x=1; for i=1:n, x=max(x,max(xs1{i})); end, x=min(x-mod(x,.1),1.0);
y=.8; for i=1:n, y=min(y,min(ys1{i})); end, y=max(y-mod(y,.1),.01);
xlim([0, x]); ylim([y, 1]); set(gca,'xtick',0:.1:1);
xlabel('Recall','FontSize',14); ylabel('Precision','FontSize',14);
end
if(~isempty(lgd1)), legend(h,lgd1,'Location','sw','FontSize',10); end
% save figure to disk (uncomment pdfcrop commands to automatically crop)
[o,~]=system('pdfcrop'); if(o==127), setenv('PATH',...
[getenv('PATH') ':/Library/TeX/texbin/:/usr/local/bin/']); end
savefig(fName1,1,'pdf','-r300','-fonts'); close(1); f1=[fName1 '.pdf'];
system(['pdfcrop -margins ''-30 -20 -50 -10 '' ' f1 ' ' f1]);
end
end
function plotBbs( res, plotName, pPage, type )
% This function plots sample fp/tp/fn bbs for given algs/exps
if(pPage==0), return; end; [nGt,nDt]=size(res);
% construct set/vid/frame index for each image
[~,setIds,vidIds,skip]=dbInfo;
k=length(res(1).gtr); is=zeros(k,3); k=0;
for s=1:length(setIds)
for v=1:length(vidIds{s})
A=loadVbb(s,v); s1=setIds(s); v1=vidIds{s}(v);
for f=skip-1:skip:A.nFrame-1, k=k+1; is(k,:)=[s1 v1 f]; end
end
end
for g=1:nGt
for d=1:nDt
% augment each bb with set/video/frame index and flatten
dtr=res(g,d).dtr; gtr=res(g,d).gtr;
for i=1:k
dtr{i}(:,7)=is(i,1); dtr{i}(:,8)=is(i,2); dtr{i}(:,9)=is(i,3);
gtr{i}(:,6)=is(i,1); gtr{i}(:,7)=is(i,2); gtr{i}(:,8)=is(i,3);
dtr{i}=dtr{i}'; gtr{i}=gtr{i}';
end
dtr=[dtr{:}]'; dtr=dtr(dtr(:,6)~=-1,:);
gtr=[gtr{:}]'; gtr=gtr(gtr(:,5)~=-1,:);
% get bb, ind, bbo, and indo according to type
if( strcmp(type,'fn') )
keep=gtr(:,5)==0; ord=randperm(sum(keep));
bbCol='r'; bboCol='y'; bbLst='-'; bboLst='--';
bb=gtr(:,1:4); ind=gtr(:,6:8); bbo=dtr(:,1:6); indo=dtr(:,7:9);
else
switch type
case 'dt', bbCol='y'; keep=dtr(:,6)>=0;
case 'fp', bbCol='r'; keep=dtr(:,6)==0;
case 'tp', bbCol='y'; keep=dtr(:,6)==1;
end
[~,ord]=sort(dtr(keep,5),'descend');
bboCol='g'; bbLst='--'; bboLst='-';
bb=dtr(:,1:6); ind=dtr(:,7:9); bbo=gtr(:,1:4); indo=gtr(:,6:8);
end
% prepare and display
n=sum(keep); bbo1=cell(1,n); O=ones(1,size(indo,1));
ind=ind(keep,:); bb=bb(keep,:); ind=ind(ord,:); bb=bb(ord,:);
for f=1:n, bbo1{f}=bbo(all(indo==ind(O*f,:),2),:); end
f=[plotName res(g,d).stre res(g,d).stra '-' type];
plotBbSheet( bb, ind, bbo1,'fName',f,'pPage',pPage,'bbCol',bbCol,...
'bbLst',bbLst,'bboCol',bboCol,'bboLst',bboLst );
end
end
end
function plotBbSheet( bb, ind, bbo, varargin )
% Draw sheet of bbs.
%
% USAGE
% plotBbSheet( R, varargin )
%
% INPUTS
% bb - [nx4] bbs to display
% ind - [nx3] the set/video/image number for each bb
% bbo - {nx1} cell of other bbs for each image (optional)
% varargin - prm struct or name/value list w following fields:
% .fName - ['REQ'] base file to save to
% .pPage - [1] num pages
% .mRows - [5] num rows / page
% .nCols - [9] num cols / page
% .scale - [2] size of image region to crop relative to bb
% .siz0 - [100 50] target size of each bb
% .pad - [4] amount of space between cells
% .bbCol - ['g'] bb color
% .bbLst - ['-'] bb LineStyle
% .bboCol - ['r'] bbo color
% .bboLst - ['--'] bbo LineStyle
dfs={'fName','REQ', 'pPage',1, 'mRows',5, 'nCols',9, 'scale',1.5, ...
'siz0',[100 50], 'pad',8, 'bbCol','g', 'bbLst','-', ...
'bboCol','r', 'bboLst','--' };
[fName,pPage,mRows,nCols,scale,siz0,pad,bbCol,bbLst, ...
bboCol,bboLst] = getPrmDflt(varargin,dfs);
n=size(ind,1); indAll=ind; bbAll=bb; bboAll=bbo;
for page=1:min(pPage,ceil(n/mRows/nCols))
Is = zeros(siz0(1)*scale,siz0(2)*scale,3,mRows*nCols,'uint8');
bbN=[]; bboN=[]; labels=repmat({''},1,mRows*nCols);
for f=1:mRows*nCols
% get fp bb (bb), double size (bb2), and other bbs (bbo)
f0=f+(page-1)*mRows*nCols; if(f0>n), break, end
[col,row]=ind2sub([nCols mRows],f);
ind=indAll(f0,:); bb=bbAll(f0,:); bbo=bboAll{f0};
hr=siz0(1)/bb(4); wr=siz0(2)/bb(3); mr=min(hr,wr);
bb2 = round(bbApply('resize',bb,scale*hr/mr,scale*wr/mr));
bbo=bbApply('intersect',bbo,bb2); bbo=bbo(bbApply('area',bbo)>0,:);
labels{f}=sprintf('%i/%i/%i',ind(1),ind(2),ind(3));
% normalize bb and bbo for siz0*scale region, then shift
bb=bbApply('shift',bb,bb2(1),bb2(2)); bb(:,1:4)=bb(:,1:4)*mr;
bbo=bbApply('shift',bbo,bb2(1),bb2(2)); bbo(:,1:4)=bbo(:,1:4)*mr;
xdel=-pad*scale-(siz0(2)+pad*2)*scale*(col-1);
ydel=-pad*scale-(siz0(1)+pad*2)*scale*(row-1);
bb=bbApply('shift',bb,xdel,ydel); bbN=[bbN; bb]; %#ok<AGROW>
bbo=bbApply('shift',bbo,xdel,ydel); bboN=[bboN; bbo]; %#ok<AGROW>
% load and crop image region
sr=seqIo(sprintf('%s/videos/set%02i/V%03i',dbInfo,ind(1),ind(2)),'r');
sr.seek(ind(3)); I=sr.getframe(); sr.close();
I=bbApply('crop',I,bb2,'replicate');
I=uint8(imResample(double(I{1}),siz0*scale));
Is(:,:,:,f)=I;
end
% now plot all and save
prm=struct('hasChn',1,'padAmt',pad*2*scale,'padEl',0,'mm',mRows,...
'showLines',0,'labels',{labels});
h=figureResized(.9,1); clf; montage2(Is,prm); hold on;
bbApply('draw',bbN,bbCol,2,bbLst); bbApply('draw',bboN,bboCol,2,bboLst);
savefig([fName int2str2(page-1,2)],h,'png','-r200','-fonts'); close(h);
if(0), save([fName int2str2(page-1,2) '.mat'],'Is'); end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function A = loadVbb( s, v )
% Load given annotation (caches AS for speed).
persistent AS pth sIds vIds; [pth1,sIds1,vIds1]=dbInfo;
if(~strcmp(pth,pth1) || ~isequal(sIds,sIds1) || ~isequal(vIds,vIds1))
[pth,sIds,vIds]=dbInfo; AS=cell(length(sIds),1e3); end
A=AS{s,v}; if(~isempty(A)), return; end
最後的結果：
