def list_to_img(all_left_img,
all_right_img,
all_left_disp,
index,
batchsize,
training=True,
use_bn=False):
list_left_img = []
list_right_img = []
list_left_disp = []
initial_index = index * batchsize
for i in range(batchsize):
left_img = Image.open(all_left_img[initial_index + i]).convert('RGB')
right_img = Image.open(all_right_img[initial_index + i]).convert('RGB')
left_disp, _ = rp.readPFM(all_left_disp[initial_index + i])
left_disp = np.ascontiguousarray(left_disp, dtype=np.float32)
if training:
w, h = left_img.size
th, tw = 256, 512
x1 = random.randint(0, w - tw)
y1 = random.randint(0, h - th)
left_img = left_img.crop((x1, y1, x1 + tw, y1 + th))
right_img = right_img.crop((x1, y1, x1 + tw, y1 + th))
left_disp = left_disp[y1:y1 + th, x1:x1 + tw]
left_img = np.array(left_img, dtype=np.float32)
right_img = np.array(right_img, dtype=np.float32)
if use_bn:
left_img = scale_crop(left_img)
right_img = scale_crop(right_img)
list_left_img.append(left_img)
list_right_img.append(right_img)
list_left_disp.append(left_disp)
batch_left_img = np.array(list_left_img)
batch_right_img = np.array(list_right_img)
batch_left_disp = np.array(list_left_disp)
return batch_left_img, batch_right_img, batch_left_disp
def disparity_loader(path):
return rp.readPFM(path)
def disparity_loader(path):
# print(path)
img = rp.readPFM(path)
plt.imshow(img)
plt.show()
return img
def disparity_loader(path):
# print(path)
return rp.readPFM(path)
def disparity_loader(path):
return rp.readPFM(path, False) #Don't flip pfm