def main(_):
model = DPQ(FLAGS)
a = "/device:GPU:0" if FLAGS.UseGPU else "/cpu:0"
print("Using device:", a, "<-", FLAGS.Device)
with tf.device(a):
queryX, queryY, db = Dataset.PreparetoEval(FLAGS.Dataset, IMAGE_WIDTH, IMAGE_HEIGHT)
result = model.Evaluate(queryX, queryY, db)
def main(_):
model = DSQ(FLAGS)
a = "/device:GPU:0" if FLAGS.UseGPU else "/cpu:0"
print("Using device:", a, "<-", FLAGS.Device)
with tf.device(a):
queryX, queryY, db = Dataset.PreparetoEval(FLAGS.Dataset, IMAGE_WIDTH,
IMAGE_HEIGHT)
fileName = model.CheckTime(queryX)
def main(_):
model = DSQ(FLAGS)
a = "/device:GPU:0" if FLAGS.UseGPU else "/cpu:0"
print("Using device:", a, "<-", FLAGS.Device)
with tf.device(a):
queryX, queryY, db = Dataset.PreparetoEval(FLAGS.Dataset, IMAGE_WIDTH, IMAGE_HEIGHT)
result = model.Evaluate(queryX, queryY, db)
now = datetime.datetime.now()
if not path.exists('results'):
os.mkdir('results')
with open('./results/result_{0}_{1}'.format(model._name, now.strftime("%Y-%m-%d %H:%M:%S")),'w') as fp:
fp.write(result + '\n')
def main(_):
model = DPQ(FLAGS)
a = "/device:GPU:0" if FLAGS.UseGPU else "/cpu:0"
print("Using device:", a, "<-", FLAGS.Device)
with tf.device(a):
queryX, queryY, db = Dataset.PreparetoEval(FLAGS.Dataset, IMAGE_WIDTH,
IMAGE_HEIGHT)
fileName = model.GetRetrievalMat(queryX, queryY, db)
if os.path.exists(fileName):
retrievalMat = np.load(fileName)
precision = np.mean(np.cumsum(retrievalMat, axis=1) /
np.arange(1, retrievalMat.shape[1] + 1, 1),
axis=0)
np.savetxt('precision.csv', precision)
totalResult = np.sum(retrievalMat, axis=1)
print(totalResult)
# [Nq, Nb] / [Nq, 1]
recall = np.mean(np.cumsum(retrievalMat, axis=1) /
totalResult[:, None],
axis=0)
np.savetxt('recall.csv', recall)
