}
batch_loss, batch_acc, batch_preds = sess.run(
[loss, accuracy, preds], feed_dict=feed_dict)
test_loss += batch_loss
test_acc += batch_acc
full_pred.extend(batch_preds)
full_batch.extend(label_batch)
test_loss /= test_batches # num_samples
test_acc /= test_batches
batchMatrix = defineImageRank(full_pred, full_batch)
classList, num_of_frames = countTry(batchMatrix, num_classes)
values = cmc_values(classList, num_of_frames)
plot_cmc(values, "CMC", num_classes)
# y_true per ogni immagine contiene il nome della classe vera
# y_pred per ogni immagine contiene il nome della classe predetta
y_true = []
y_pred = []
for i in range(0, len(full_batch)):
y_true.append(batchMatrix[i][num_classes]) #classe vera
y_pred.append(batchMatrix[i][0]) #classe predetta
class_rep = classification_report(y_true, y_pred)
class_names = []
for i in range(0, num_classes):
class_names.append(i)
cnf_matrix = confusion_matrix(y_true, y_pred)
[
dist_matrix.append(
euclidean_distance(np.array(img_feature),
np.array(gallery_features),
gallery_true_class, num_classes,
frames_gallery))
for img_feature in batch_features
]
dist_min = dist_matrix_avg(dist_matrix, frames_gallery)
#dist_min = dist_matrix_min(dist_matrix, frames_gallery)
batchMatrix = defineImageRank1(dist_min, full_label)
classList, num_of_frames = countTry(batchMatrix, num_classes_test)
values = cmc_values(classList, num_of_frames)
plot_cmc(values, "CMC", num_classes_test)
# y_true per ogni immagine contiene il nome della classe vera
# y_pred per ogni immagine contiene il nome della classe predetta
y_true = []
y_pred = []
for i in range(0, len(full_label)):
y_true.append(batchMatrix[i][num_classes_test]) # classe vera
y_pred.append(batchMatrix[i][0]) # classe predetta
test_acc = accuracy_score(y_true, y_pred)
class_names = []
for i in range(0, num_classes):
class_names.append(i)
#confusion matrix here