def test(path):
train_dir = pj(path, 'training')
check_path = tf.train.latest_checkpoint(train_dir)
print 'Restoring from:', check_path
net = NetClf(check_path, gpu)
data = ut.load(pj(path, 'test.pk'))
labels = []
probs = []
accs = []
for i in xrange(len(data)):
ex = data[i]
label = ex['is_gripping']
ex = {k: ig.uncompress(ex[k]) for k in im_names}
prob = net.predict(**ex)
print prob, label
pred = int(prob >= 0.5)
labels.append(label)
probs.append(prob)
accs.append(pred == label)
labels = np.array(labels, 'bool')
probs = np.array(probs, 'float32')
accs = np.array(accs)
print 'Accuracy:', np.mean(accs)
print 'mAP:', sklearn.metrics.average_precision_score(labels, probs)
def im(x, crop=False, compress=True):
x = ig.uncompress(x)
x = np.array(x)
if crop:
x = crop_kinect(x)
#ig.show(x)
x = ig.scale(x, (256, 256), 1)
if compress:
x = ig.compress(x)
return x
def load_im(k, v):
if k.startswith('gel') or k.startswith('im'):
im = ig.uncompress(v)
elif k.startswith('depth'):
#v = np.tile(v, (1, 1, 3))
im = v.astype('float32')
else:
raise RuntimeError()
if center_crop:
im = ut.crop_center(im, 224)
return im
def vis_example(db_file):
with h5py.File(db_file, 'r') as db:
pre, mid, post = milestone_frames(db)
#sc = lambda x : ig.scale(x, (600, None))
sc = lambda x: x
crop_kinect = lambda x: x
im_base = ig.uncompress(db['color_image_KinectA'][mid])
im_mid = sc(crop_kinect(im_base))
im_post = sc(
crop_kinect(ig.uncompress(db['color_image_KinectA'][post])))
im_crop = sc(
crop_obj(im_base,
np.array(db['cylinder_data']['xyz_kinect'].value)))
depth = sc(color_depth(crop_kinect(db['depth_image_KinectA'][mid])))
gel_a_0 = sc(ig.uncompress(db['GelSightA_image'][pre]))
gel_b_0 = sc(ig.uncompress(db['GelSightB_image'][pre]))
gel_a_1 = sc(ig.uncompress(db['GelSightA_image'][mid]))
gel_b_1 = sc(ig.uncompress(db['GelSightB_image'][mid]))
row = [
'Color:', im_mid, 'Depth:', depth, 'Gel_A_1:', gel_a_1, 'Gel_B_1:',
gel_b_1, 'Gel_A_0:', gel_a_0, 'Gel_B_0:', gel_b_0, 'Im after:',
im_post, 'Cropped:', im_crop, 'Name:',
str(np.array(db['object_name'].value[0])), 'Path:',
db_file.split('/')[-1]
]
return row
def train_clf(pr):
data = ut.load(pj(pr.dsdir, 'train.pk'))
xs, ys = [], []
for ex in data:
ex = copy.copy(ex)
for k, v in ex.items():
if k.startswith('gel'):
ex[k] = ut.crop_center(ig.uncompress(v), 224)
xs.append(example_feats(ex, pr))
ys.append(ex['is_gripping'])
xs = np.array(xs, 'float32')
ys = np.array(ys, 'int64')
clf = sklearn.pipeline.Pipeline([
('scale',
sklearn.preprocessing.StandardScaler(with_mean=True, with_std=True)),
('svm', sklearn.svm.SVC(C=1., kernel='linear'))
])
clf.fit(xs, ys)
ut.save(pj(pr.resdir, 'clf.pk'), clf)
def load_im(k, v):
if k.startswith('gel') or k.startswith('im'):
im = ig.uncompress(v)
elif k.startswith('depth'):
#v = np.tile(v, (1, 1, 3))
im = v.astype('float32')
else:
raise RuntimeError()
if crop_type == 'center':
crops = [ut.crop_center(im, 224)]
elif crop_type == 'multi':
crops = []
dh = (im.shape[0] - crop_dim)
num_dim_samples = 3
for y in np.linspace(0, dh, num_dim_samples).astype('l'):
dw = (im.shape[1] - crop_dim)
for x in np.linspace(0, dw, num_dim_samples).astype('l'):
crops.append(im[y:y + crop_dim, x:x + crop_dim])
return ut.shuffled_with_seed(crops, k.split('_')[0] + str(i))