From 925176b50390c5eba974d3609abc203527ae8ba6 Mon Sep 17 00:00:00 2001 From: Vasil Zlatanov Date: Wed, 5 Dec 2018 18:23:40 +0000 Subject: Move to lib folder and fix kmeans --- lib/rerank.py | 82 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 82 insertions(+) create mode 100644 lib/rerank.py (limited to 'lib/rerank.py') diff --git a/lib/rerank.py b/lib/rerank.py new file mode 100644 index 0000000..6b20f53 --- /dev/null +++ b/lib/rerank.py @@ -0,0 +1,82 @@ +from scipy.spatial.distance import cdist +import numpy as np + +def re_ranking(probFea,galFea,k1,k2,lambda_value, MemorySave = False, Minibatch = 2000): + + query_num = probFea.shape[0] + all_num = query_num + galFea.shape[0] + feat = np.append(probFea,galFea,axis = 0) + feat = feat.astype(np.float16) + print('computing original distance') + if MemorySave: + original_dist = np.zeros(shape = [all_num,all_num],dtype = np.float16) + i = 0 + while True: + it = i + Minibatch + if it < np.shape(feat)[0]: + original_dist[i:it,] = np.power(cdist(feat[i:it,],feat),2).astype(np.float16) + else: + original_dist[i:,:] = np.power(cdist(feat[i:,],feat),2).astype(np.float16) + break + i = it + else: + original_dist = cdist(feat,feat).astype(np.float16) + original_dist = np.power(original_dist,2).astype(np.float16) + del feat + gallery_num = original_dist.shape[0] + original_dist = np.transpose(original_dist/np.max(original_dist,axis = 0)) + V = np.zeros_like(original_dist).astype(np.float16) + initial_rank = np.argsort(original_dist).astype(np.int32) + + + print('starting re_ranking') + for i in range(all_num): + # k-reciprocal neighbors + forward_k_neigh_index = initial_rank[i,:k1+1] + backward_k_neigh_index = initial_rank[forward_k_neigh_index,:k1+1] + fi = np.where(backward_k_neigh_index==i)[0] + k_reciprocal_index = forward_k_neigh_index[fi] + k_reciprocal_expansion_index = k_reciprocal_index + for j in range(len(k_reciprocal_index)): + candidate = k_reciprocal_index[j] + candidate_forward_k_neigh_index = initial_rank[candidate,:int(np.around(k1/2))+1] + candidate_backward_k_neigh_index = initial_rank[candidate_forward_k_neigh_index,:int(np.around(k1/2))+1] + fi_candidate = np.where(candidate_backward_k_neigh_index == candidate)[0] + candidate_k_reciprocal_index = candidate_forward_k_neigh_index[fi_candidate] + if len(np.intersect1d(candidate_k_reciprocal_index,k_reciprocal_index))> 2/3*len(candidate_k_reciprocal_index): + k_reciprocal_expansion_index = np.append(k_reciprocal_expansion_index,candidate_k_reciprocal_index) + + k_reciprocal_expansion_index = np.unique(k_reciprocal_expansion_index) + weight = np.exp(-original_dist[i,k_reciprocal_expansion_index]) + V[i,k_reciprocal_expansion_index] = weight/np.sum(weight) + original_dist = original_dist[:query_num,] + if k2 != 1: + V_qe = np.zeros_like(V,dtype=np.float16) + for i in range(all_num): + V_qe[i,:] = np.mean(V[initial_rank[i,:k2],:],axis=0) + V = V_qe + del V_qe + del initial_rank + invIndex = [] + for i in range(gallery_num): + invIndex.append(np.where(V[:,i] != 0)[0]) + + jaccard_dist = np.zeros_like(original_dist,dtype = np.float16) + + for i in range(query_num): + temp_min = np.zeros(shape=[1,gallery_num],dtype=np.float16) + indNonZero = np.where(V[i,:] != 0)[0] + indImages = [] + indImages = [invIndex[ind] for ind in indNonZero] + for j in range(len(indNonZero)): + temp_min[0,indImages[j]] = temp_min[0,indImages[j]]+ np.minimum(V[i,indNonZero[j]],V[indImages[j],indNonZero[j]]) + jaccard_dist[i] = 1-temp_min/(2-temp_min) + + final_dist = jaccard_dist*(1-lambda_value) + original_dist*lambda_value + del original_dist + del V + del jaccard_dist + final_dist = final_dist[:query_num,query_num:] + + return final_dist + -- cgit v1.2.3-54-g00ecf From 689ab1f6e0f9923d06507d6835f2d0d068a83d22 Mon Sep 17 00:00:00 2001 From: Vasil Zlatanov Date: Wed, 5 Dec 2018 18:39:52 +0000 Subject: Replace verbose with logging --- evaluate.py | 15 +++++++++------ lib/kmean.py | 2 ++ lib/rerank.py | 5 +++-- 3 files changed, 14 insertions(+), 8 deletions(-) (limited to 'lib/rerank.py') diff --git a/evaluate.py b/evaluate.py index ace647f..76ee472 100755 --- a/evaluate.py +++ b/evaluate.py @@ -31,6 +31,8 @@ from scipy.spatial.distance import cdist sys.path.append('lib') from rerank import re_ranking from kmean import create_kmean_clusters +import logging +from logging import debug parser = argparse.ArgumentParser() parser.add_argument("-t", "--train", help="Use test data instead of query", action='store_true') @@ -53,10 +55,8 @@ parser.add_argument("--data", help="Data folder with features data", default='da args = parser.parse_args() - -def verbose(*text): - if args.verbose: - print(text) +if args.verbose: + logging.basicConfig(level=logging.DEBUG) def draw_results(test_label, pred_label): acc_sc = accuracy_score(test_label, pred_label) @@ -72,8 +72,8 @@ def draw_results(test_label, pred_label): def test_model(gallery_data, probe_data, gallery_label, probe_label, gallery_cam, probe_cam, showfiles_train, showfiles_test, args): - verbose("probe shape:", probe_data.shape) - verbose("gallery shape:", gallery_data.shape) + debug("probe shape:", probe_data.shape) + debug("gallery shape:", gallery_data.shape) if args.rerank: distances = re_ranking(probe_data, gallery_data, @@ -137,6 +137,7 @@ def test_model(gallery_data, probe_data, gallery_label, probe_label, gallery_cam return target_pred def main(): + logging.debug("Verbose mode is on") mat = scipy.io.loadmat(os.path.join(args.data,'cuhk03_new_protocol_config_labeled.mat')) camId = mat['camId'] filelist = mat['filelist'] @@ -167,9 +168,11 @@ def main(): test_table = np.arange(1, args.multrank+1) if (args.normalise): + debug("Normalising data") train_data = np.divide(train_data,LA.norm(train_data, axis=0)) test_data = np.divide(test_data, LA.norm(test_data, axis=0)) if(args.kmean): + debug("Using Kmeans") train_data, train_label, train_cam = create_kmean_clusters(feature_vectors, labels, gallery_idx, diff --git a/lib/kmean.py b/lib/kmean.py index 58309bc..f041b19 100644 --- a/lib/kmean.py +++ b/lib/kmean.py @@ -1,3 +1,4 @@ +from logging import debug import numpy as np from sklearn.cluster import KMeans @@ -39,6 +40,7 @@ def create_kmean_clusters(feature_vectors, labels, gallery_idx, camId): cllab.append(km_labels[i][j]) train_data = np.array(cl) + debug("Kmean data has shape", train_data.shape) train_label = np.array(cllab) train_cam = np.array([clustercam[i] for i in range(2)]).reshape(train_label.shape[0]) return train_data, train_label, train_cam diff --git a/lib/rerank.py b/lib/rerank.py index 6b20f53..6fb5b7b 100644 --- a/lib/rerank.py +++ b/lib/rerank.py @@ -1,5 +1,6 @@ from scipy.spatial.distance import cdist import numpy as np +from logging import debug def re_ranking(probFea,galFea,k1,k2,lambda_value, MemorySave = False, Minibatch = 2000): @@ -7,7 +8,7 @@ def re_ranking(probFea,galFea,k1,k2,lambda_value, MemorySave = False, Minibatch all_num = query_num + galFea.shape[0] feat = np.append(probFea,galFea,axis = 0) feat = feat.astype(np.float16) - print('computing original distance') + debug('computing original distance') if MemorySave: original_dist = np.zeros(shape = [all_num,all_num],dtype = np.float16) i = 0 @@ -29,7 +30,7 @@ def re_ranking(probFea,galFea,k1,k2,lambda_value, MemorySave = False, Minibatch initial_rank = np.argsort(original_dist).astype(np.int32) - print('starting re_ranking') + debug('starting re_ranking') for i in range(all_num): # k-reciprocal neighbors forward_k_neigh_index = initial_rank[i,:k1+1] -- cgit v1.2.3-54-g00ecf