jts = poseNet.computeOutput([test_data1, test_data12, test_data14])
for i in xrange(test_data1.shape[0]):
joints.append(jts[i].reshape(1, 3) *
(testSeqs[0].config['cube'][2] / 2.) +
testSeqs[0].data[i].com)
gt3D.extend([j.gt3Dorig[13].reshape(1, 3) for j in testSeqs[1].data])
jts = poseNet.computeOutput([test_data2, test_data22, test_data24])
for i in range(test_data2.shape[0]):
joints.append(jts[i].reshape(1, 3) *
(testSeqs[1].config['cube'][2] / 2.) +
testSeqs[1].data[i].com)
hpe = NYUHandposeEvaluation(gt3D, joints)
hpe.subfolder += '/' + eval_prefix + '/'
mean_error = hpe.getMeanError()
max_error = hpe.getMaxError()
print("Mean error: {}mm, max error: {}mm".format(mean_error, max_error))
# save results
cPickle.dump(joints,
open(
"./eval/{}/result_{}_{}.pkl".format(
eval_prefix,
os.path.split(__file__)[1], eval_prefix), "wb"),
protocol=cPickle.HIGHEST_PROTOCOL)
print "Testing baseline"
#################################
# BASELINE
test_data, _ = testDataSet.imgStackDepthOnly(seq.name)
jts_embed = poseNet.computeOutput(test_data)
# Backtransform from embedding
# jts = pca.inverse_transform(jts_embed)
jts = jts_embed
for i in xrange(test_data.shape[0]):
joints.append(jts[i].reshape((-1, 3)) *
(seq.config['cube'][2] / 2.) + seq.data[i].com)
joints = numpy.array(joints)
hpe = NYUHandposeEvaluation(gt3D, joints)
hpe.subfolder += '/' + eval_prefix + '/'
print("Train samples: {}, test samples: {}".format(train_data.shape[0],
len(gt3D)))
print("Mean error: {}mm, max error: {}mm".format(hpe.getMeanError(),
hpe.getMaxError()))
print("{}".format(
[hpe.getJointMeanError(j) for j in range(joints[0].shape[0])]))
print("{}".format(
[hpe.getJointMaxError(j) for j in range(joints[0].shape[0])]))
# save results
cPickle.dump(joints,
open(
"./eval/{}/result_{}_{}.pkl".format(
eval_prefix,
os.path.split(__file__)[1], eval_prefix), "wb"),
protocol=cPickle.HIGHEST_PROTOCOL)
print "Testing baseline"
hpe[-1].plotEvaluation('ablation',
methodName='Final',
baseline=plot_list[:-1])
else:
# plot list
plot_list = []
# Comparison
# siggraph
data_siggrpah = di.loadBaseline(
'../dataset/NYU/test/test_predictions.mat', np.asarray(gt3D))
hpe_siggrpah = NYUHandposeEvaluation(gt3D, data_siggrpah)
hpe_siggrpah.subfolder += 'comparison/'
print("Tompson et al. Siggraph 2014")
print("Mean error: {}mm".format(hpe_siggrpah.getMeanError()))
plot_list.append(('HeatMap', hpe_siggrpah))
# cvww15 deeppriori
data_deeppriori = di.loadBaseline(
'../result/CVWW15/CVWW15_NYU_Prior-Refinement.txt')
hpe_deeppriori = NYUHandposeEvaluation(gt3D, data_deeppriori)
hpe_deeppriori.subfolder += 'comparison/'
print("Oberweger et al. CVWW 2015")
print("Mean error: {}mm".format(hpe_deeppriori.getMeanError()))
plot_list.append(('DeepPrior', hpe_deeppriori))
# iccv15 deep loopback
data_loopback = di.loadBaseline(
'../result/ICCV15/ICCV15_NYU_Feedback.txt')
hpe_loopback = NYUHandposeEvaluation(gt3D, data_loopback)