- Timestamp:
- Feb 17, 2006, 5:07:36 PM (19 years ago)
- Branches:
- feature/autosink, feature/cnn, feature/cnn_org, feature/constantq, feature/crepe, feature/crepe_org, feature/pitchshift, feature/pydocstrings, feature/timestretch, fix/ffmpeg5, master, pitchshift, sampler, timestretch, yinfft+
- Children:
- c912c67
- Parents:
- 677b267
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
python/test/bench/onset/bench-onset
r677b267 re968939 4 4 from aubio.tasks import * 5 5 6 7 8 9 def mmean(l): 10 return sum(l)/float(len(l)) 11 12 def stdev(l): 13 smean = 0 14 lmean = mmean(l) 15 for i in l: 16 smean += (i-lmean)**2 17 smean *= 1. / len(l) 18 return smean**.5 19 6 20 class benchonset(bench): 21 22 valuenames = ['orig','missed','Tm','expc','bad','Td'] 23 valuelists = ['l','labs'] 24 printnames = [ 'mode', 'thres', 'dist', 'prec', 'recl', 'Ttrue', 'Tfp', 'Tfn', 'Tm', 'Td', 25 'aTtrue', 'aTfp', 'aTfn', 'aTm', 'aTd', 'mean', 'smean', 'amean', 'samean'] 26 27 formats = {'mode': "%12s" , 28 'thres': "%5.4s", 29 'dist': "%5.4s", 30 'prec': "%5.4s", 31 'recl': "%5.4s", 32 33 'Ttrue': "%5.4s", 34 'Tfp': "%5.4s", 35 'Tfn': "%5.4s", 36 'Tm': "%5.4s", 37 'Td': "%5.4s", 38 39 'aTtrue':"%5.4s", 40 'aTfp': "%5.4s", 41 'aTfn': "%5.4s", 42 'aTm': "%5.4s", 43 'aTd': "%5.4s", 44 45 'mean': "%5.40s", 46 'smean': "%5.40s", 47 'amean': "%5.40s", 48 'samean': "%5.40s"} 7 49 8 def dir_eval(self): 9 self.P = 100*float(self.expc-self.missed-self.merged)/(self.expc-self.missed-self.merged + self.bad+self.doubled) 10 self.R = 100*float(self.expc-self.missed-self.merged)/(self.expc-self.missed-self.merged + self.missed+self.merged) 11 if self.R < 0: self.R = 0 12 self.F = 2* self.P*self.R / (self.P+self.R) 13 14 self.values = [self.params.onsetmode, 15 "%2.3f" % self.params.threshold, 16 self.orig, 17 self.expc, 18 self.missed, 19 self.merged, 20 self.bad, 21 self.doubled, 22 (self.orig-self.missed-self.merged), 23 "%2.3f" % (100*float(self.orig-self.missed-self.merged)/(self.orig)), 24 "%2.3f" % (100*float(self.bad+self.doubled)/(self.orig)), 25 "%2.3f" % (100*float(self.orig-self.missed)/(self.orig)), 26 "%2.3f" % (100*float(self.bad)/(self.orig)), 27 "%2.3f" % self.P, 28 "%2.3f" % self.R, 29 "%2.3f" % self.F ] 50 def file_gettruth(self,input): 51 from os.path import isfile 52 ftrulist = [] 53 # search for match as filetask.input,".txt" 54 ftru = '.'.join(input.split('.')[:-1]) 55 ftru = '.'.join((ftru,'txt')) 56 if isfile(ftru): 57 ftrulist.append(ftru) 58 else: 59 # search for matches for filetask.input in the list of results 60 for i in range(len(self.reslist)): 61 check = '.'.join(self.reslist[i].split('.')[:-1]) 62 check = '_'.join(check.split('_')[:-1]) 63 if check == '.'.join(input.split('.')[:-1]): 64 ftrulist.append(self.reslist[i]) 65 return ftrulist 30 66 31 67 def file_exec(self,input,output): 32 68 filetask = self.task(input,params=self.params) 33 69 computed_data = filetask.compute_all() 34 results = filetask.eval(computed_data) 35 self.orig += filetask.orig 36 self.missed += filetask.missed 37 self.merged += filetask.merged 38 self.expc += filetask.expc 39 self.bad += filetask.bad 40 self.doubled += filetask.doubled 41 70 ftrulist = self.file_gettruth(filetask.input) 71 for i in ftrulist: 72 #print i 73 filetask.eval(computed_data,i,mode='rocloc',vmode='') 74 for i in self.valuenames: 75 self.v[i] += filetask.v[i] 76 for i in filetask.v['l']: 77 self.v['l'].append(i) 78 for i in filetask.v['labs']: 79 self.v['labs'].append(i) 80 81 def dir_exec(self): 82 """ run file_exec on every input file """ 83 self.l , self.labs = [], [] 84 self.v = {} 85 for i in self.valuenames: 86 self.v[i] = 0. 87 for i in self.valuelists: 88 self.v[i] = [] 89 self.v['thres'] = self.params.threshold 90 act_on_files(self.file_exec,self.sndlist,self.reslist, \ 91 suffix='',filter=sndfile_filter) 92 93 def dir_eval(self): 94 totaltrue = self.v['expc']-self.v['bad']-self.v['Td'] 95 totalfp = self.v['bad']+self.v['Td'] 96 totalfn = self.v['missed']+self.v['Tm'] 97 self.P = 100*float(totaltrue)/max(totaltrue + totalfp,1) 98 self.R = 100*float(totaltrue)/max(totaltrue + totalfn,1) 99 if self.R < 0: self.R = 0 100 self.F = 2.* self.P*self.R / max(float(self.P+self.R),1) 101 102 N = float(len(self.reslist)) 103 104 self.v['mode'] = self.params.onsetmode 105 self.v['thres'] = "%2.3f" % self.params.threshold 106 self.v['dist'] = "%2.3f" % self.F 107 self.v['prec'] = "%2.3f" % self.P 108 self.v['recl'] = "%2.3f" % self.R 109 self.v['Ttrue'] = totaltrue 110 self.v['Tfp'] = totalfp 111 self.v['Tfn'] = totalfn 112 self.v['aTtrue'] = totaltrue/N 113 self.v['aTfp'] = totalfp/N 114 self.v['aTfn'] = totalfn/N 115 self.v['aTm'] = self.v['Tm']/N 116 self.v['aTd'] = self.v['Td']/N 117 self.v['mean'] = mmean(self.v['l']) 118 self.v['smean'] = stdev(self.v['l']) 119 self.v['amean'] = mmean(self.v['labs']) 120 self.v['samean'] = stdev(self.v['labs']) 42 121 43 122 def run_bench(self,modes=['dual'],thresholds=[0.5]): … … 45 124 self.thresholds = thresholds 46 125 47 self.pretty_ print(self.titles)126 self.pretty_titles() 48 127 for mode in self.modes: 49 128 self.params.onsetmode = mode … … 52 131 self.dir_exec() 53 132 self.dir_eval() 54 self.pretty_print(self.values) 133 self.pretty_print() 134 #print self.v 135 136 def pretty_print(self,sep='|'): 137 for i in self.printnames: 138 print self.formats[i] % self.v[i], sep, 139 print 140 141 def pretty_titles(self,sep='|'): 142 for i in self.printnames: 143 print self.formats[i] % i, sep, 144 print 55 145 56 146 def auto_learn(self,modes=['dual'],thresholds=[0.1,1.5]): 57 147 """ simple dichotomia like algorithm to optimise threshold """ 58 148 self.modes = modes 59 self.pretty_ print(self.titles)149 self.pretty_titles() 60 150 for mode in self.modes: 61 151 steps = 10 … … 67 157 self.dir_exec() 68 158 self.dir_eval() 69 self.pretty_print( self.values)159 self.pretty_print() 70 160 topF = self.F 71 161 … … 73 163 self.dir_exec() 74 164 self.dir_eval() 75 self.pretty_print( self.values)165 self.pretty_print() 76 166 lessF = self.F 77 167 … … 80 170 self.dir_exec() 81 171 self.dir_eval() 82 self.pretty_print( self.values)172 self.pretty_print() 83 173 if self.F == 100.0 or self.F == topF: 84 174 print "assuming we converged, stopping" … … 98 188 lesst /= 2. 99 189 100 def auto_learn2(self,modes=['dual'],thresholds=[0. 1,1.0]):190 def auto_learn2(self,modes=['dual'],thresholds=[0.00001,1.0]): 101 191 """ simple dichotomia like algorithm to optimise threshold """ 102 192 self.modes = modes 103 self.pretty_ print(self.titles)193 self.pretty_titles([]) 104 194 for mode in self.modes: 105 195 steps = 10 106 step = thresholds[1] 107 curt = thresholds[0] 196 step = 0.4 108 197 self.params.onsetmode = mode 109 110 self.params.threshold = curt 111 self.dir_exec() 112 self.dir_eval() 113 self.pretty_print(self.values) 114 curexp = self.expc 198 self.params.threshold = thresholds[0] 199 cur = 0 115 200 116 201 for i in range(steps): 117 if curexp < self.orig:118 #print "we found at most less onsets than annotated"119 self.params.threshold -= step120 step /= 2121 elif curexp > self.orig:122 #print "we found more onsets than annotated"123 self.params.threshold += step124 step /= 2125 202 self.dir_exec() 126 203 self.dir_eval() 127 curexp = self.expc 128 self.pretty_print(self.values) 129 if self.orig == 100.0 or self.orig == self.expc: 130 print "assuming we converged, stopping" 204 self.pretty_print() 205 new = self.P 206 if self.R == 0.0: 207 #print "Found maximum, highering" 208 step /= 2. 209 self.params.threshold -= step 210 elif new == 100.0: 211 #print "Found maximum, highering" 212 step *= .99 213 self.params.threshold += step 214 elif cur > new: 215 #print "lower" 216 step /= 2. 217 self.params.threshold -= step 218 elif cur < new: 219 #print "higher" 220 step *= .99 221 self.params.threshold += step 222 else: 223 print "Assuming we converged" 131 224 break 225 cur = new 226 132 227 133 228 if __name__ == "__main__": … … 136 231 else: print "ERR: a path is required"; sys.exit(1) 137 232 modes = ['complex', 'energy', 'phase', 'specdiff', 'kl', 'mkl', 'dual'] 138 #modes = [ ' complex' ]139 thresholds = [ 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5 ]233 #modes = [ 'phase' ] 234 thresholds = [ 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2] 140 235 #thresholds = [1.5] 141 236 … … 146 241 benchonset.params = taskparams() 147 242 benchonset.task = taskonset 148 149 benchonset.titles = [ 'mode', 'thres', 'orig', 'expc', 'missd', 'mergd', 150 'bad', 'doubl', 'corrt', 'GD', 'FP', 'GD-merged', 'FP-pruned', 151 'prec', 'recl', 'dist' ] 152 benchonset.formats = ["%12s" , "| %6s", "| %6s", "| %6s", "| %6s", "| %6s", 153 "| %6s", "| %6s", "| %6s", "| %8s", "| %8s", "| %8s", "| %8s", 154 "| %6s", "| %6s", "| %6s"] 243 benchonset.valuesdict = {} 244 155 245 156 246 try: 157 benchonset.auto_learn2(modes=modes)158 #benchonset.run_bench(modes=modes)247 #benchonset.auto_learn2(modes=modes) 248 benchonset.run_bench(modes=modes,thresholds=thresholds) 159 249 except KeyboardInterrupt: 160 250 sys.exit(1)
Note: See TracChangeset
for help on using the changeset viewer.