def gist_train_good_svms(all_settings, d_train): gist_table = np.load(config.get_gist_dict_filename(d_train.name)) for sett_idx in range(comm_rank, len(all_settings), comm_size): sett = all_settings[sett_idx] cls = sett[0] C = sett[1] kernel = sett[2] gamma = sett[3] gist = GistClassifier(cls, d_train, gist_table) filename = config.get_gist_crossval_filename(d_train, cls) gist.train_svm(d_train, kernel, C, gamma)
def read_best_svms_from_file(d_train): all_settings = [] for cls in config.pascal_classes: filename = config.get_gist_crossval_filename(d_train, cls) lines = open(filename,'r').readlines() best_ap = 0 best_line_idx = -1 for line_idx, line in enumerate(lines): ap = float(line.split()[-1]) if ap > best_ap: best_ap = ap best_line_idx = line_idx best_line = lines[best_line_idx].split() kernel = best_line[0] C = int(best_line[1].split('=')[1]) gamma = float(best_line[2].split('=')[1]) all_settings.append([cls, C, kernel, gamma, best_ap]) return all_settings
def gist_evaluate_svms(d_train, d_val): gist_scores = np.zeros((len(d_val.images), len(d_val.classes))) gist_table = np.load(config.get_gist_dict_filename(d_train.name)) kernels = ['rbf', 'linear', 'poly'] Cs = [1,10,100] gammas = [0,0.3,1] setts = list(itertools.product(kernels, Cs, gammas)) val_gt = d_val.get_cls_ground_truth() for cls_idx in range(len(d_val.classes)): cls = d_val.classes[cls_idx] gist = GistClassifier(cls, d_train, gist_table=gist_table, d_val=d_val) filename = config.get_gist_crossval_filename(d_train, cls) # doing some crossval right here!!! for set_idx in range(comm_rank, len(setts), comm_size): sett = setts[len(setts)-1-set_idx] kernel = sett[0] C = sett[1] gamma = sett[2] train_ap = gist.train_svm(d_train, kernel, C, gamma) val_gist_table = np.load(config.get_gist_dict_filename(d_val.name)) gist_scores = svm_proba(val_gist_table, gist.svm)[:,1] val_ap,_,_ = Evaluation.compute_cls_pr(gist_scores, val_gt.subset_arr(cls)) w = open(filename, 'a') w.write('%s C=%d gamma=%f - train: %f, val: %f\n'%(kernel, C, gamma, train_ap, val_ap)) w.close() print 'ap on val: %f'%val_ap print '%d at safebarrier'%comm_rank safebarrier(comm) gist_scores = comm.reduce(gist_scores) if comm_rank == 0: print gist_scores filename = config.get_gist_classifications_filename(d_val) cPickle.dump(gist_scores, open(filename,'w')) res = Evaluation.compute_cls_pr(gist_scores, val_gt.arr) print res