def show_pose(dataset_model, dataset_image, base_dir, outputs, list_file,
save_dir, is_flip, gif):
if list_file is None:
names = util.load_names(dataset_image)
else:
with open(list_file) as f:
names = [line.strip() for line in f]
assert len(names) == outputs.shape[0]
for idx, (name, pose) in enumerate(zip(names, outputs)):
img = util.load_image(dataset_image,
os.path.join(base_dir, name),
is_flip=is_flip)
img = img.astype(np.float32)
img = (img - img.min()) / (img.max() - img.min()) * 255
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
img = util.draw_pose(dataset_model, img, pose)
cv2.imshow('result', img / 255)
if save_dir is not None:
cv2.imwrite(os.path.join(save_dir, '{:>06d}.png'.format(idx)), img)
ch = cv2.waitKey(25)
if ch == ord('q'):
break
if gif and save_dir is not None:
os.system(
'convert -loop 0 -page +0+0 -delay 25 {0}/*.png {0}/output.gif'.
format(save_dir))
def main():
if len(sys.argv) < 4:
print_usage()
dataset = sys.argv[1]
base_dir = sys.argv[2]
out_file = sys.argv[3]
phase = sys.argv[4]
names = util.load_names(dataset, phase)
lbls = util.load_labels(dataset, phase)
centers = []
for idx, strin in enumerate(lbls):
# load label data
joints = np.asarray(np.reshape(strin.split(), (21, 3)),
dtype=np.float32)
# convert label data from world coordinates to pixel locations
joints, skel_camcoords = util.world2pixel(joints, dataset)
# calculate centers
c = util.get_center_fpad(joints)
c = np.asarray(c, dtype=np.float32)
centers.append(c.reshape((1, 3)))
if idx % 500 == 0:
print('{}/{}'.format(idx + 1, len(names)))
util.save_results(centers, out_file)
def main():
if len(sys.argv) < 3:
print_usage()
dataset = sys.argv[1]
out_file = sys.argv[2]
data_dir_dict = {
'nyu': config.nyu_data_dir,
'icvl': config.icvl_data_dir,
'msra': config.msra_data_dir,
'hands17': config.hands_data_dir
}
base_dir = data_dir_dict[dataset] #sys.argv[3]
batch_size = 64
if len(sys.argv) == 4:
batch_size = int(sys.argv[3])
# generate deploy prototxt
make_baseline_net(
os.path.join('/home/boonyew/Documents/PoseRen', 'models/'), dataset)
make_pose_ren_net(
os.path.join('/home/boonyew/Documents/PoseRen', 'models/'), dataset)
hand_model = ModelPoseREN(dataset)
names = util.load_names(dataset)
centers = util.load_centers(dataset)
if dataset == 'msra':
# the last index of frames belong to the same subject
msra_id_split_range = np.array(
[8499, 16991, 25403, 33891, 42391, 50888, 59385, 67883, 76375]) - 1
results = []
for test_id in range(9):
hand_model.reset_model(dataset, test_id)
sidx = msra_id_split_range[test_id - 1] + 1 if test_id else 0
eidx = msra_id_split_range[test_id]
sub_names = names[sidx:eidx + 1]
sub_centers = centers[sidx:eidx + 1]
print('evaluating for subject {} ...'.format(test_id))
sub_results = hand_model.detect_files(base_dir,
sub_names,
sub_centers,
max_batch=batch_size)
if test_id == 0:
results = sub_results
else:
results = np.concatenate((results, sub_results), axis=0)
util.save_results(results, out_file)
else:
results = hand_model.detect_files(base_dir,
names,
centers,
max_batch=batch_size)
util.save_results(results, out_file)
def main():
if len(sys.argv) < 5:
print_usage()
dataset = sys.argv[1]
model = sys.argv[2]
out_file = sys.argv[3]
base_dir = sys.argv[4]
hand_model = HandModel(dataset, model)
names = util.load_names(dataset)
centers = util.load_centers(dataset)
results = hand_model.detect_files(base_dir, names, centers)
save_results(results, out_file)
def main():
if len(sys.argv) < 3:
print_usage()
dataset = sys.argv[1]
out_file = sys.argv[2]
data_dir_dict = {'nyu': config.nyu_data_dir,
'icvl': config.icvl_data_dir + 'test/Depth/',
'msra': config.msra_data_dir}
base_dir = data_dir_dict[dataset] #sys.argv[3]
batch_size = 64
if len(sys.argv) == 4:
batch_size = int(sys.argv[3])
# generate deploy prototxt
make_baseline_net(os.path.join(ROOT_DIR, '../models'), dataset)
make_pose_ren_net(os.path.join(ROOT_DIR, '../models'), dataset)
hand_model = ModelPoseREN(dataset)
names = util.load_names(dataset)
centers = util.load_centers(dataset)
# print("centers = {}".format(np.shape(centers)))
# c = centers[0]
# print("centers[0] = {}".format(list(c)))
if dataset == 'msra':
# the last index of frames belong to the same subject
msra_id_split_range = np.array([8499, 16991, 25403, 33891, 42391, 50888, 59385, 67883, 76375]) - 1
results = []
for test_id in xrange(9):
hand_model.reset_model(dataset, test_id)
sidx = msra_id_split_range[test_id-1] + 1 if test_id else 0
eidx = msra_id_split_range[test_id]
sub_names = names[sidx:eidx+1]
sub_centers= centers[sidx:eidx + 1]
print('evaluating for subject {} ...'.format(test_id))
sub_results = hand_model.detect_files(base_dir, sub_names, sub_centers, max_batch=batch_size)
if test_id == 0:
results = sub_results
else:
results = np.concatenate((results, sub_results), axis=0)
util.save_results(results, out_file)
else:
results = hand_model.detect_files(base_dir, names, centers, max_batch=batch_size)
print(results[0])
img = cv2.imread("/media/reborn/Others/Study/Reborn/Github/Pose-REN/test/image_0000.png", 2).astype("float32")
img_show = show_results(img, results[0], dataset)
cv2.imwrite('result.jpg', img_show)
util.save_results(results, out_file)
def main():
if len(sys.argv) < 4:
print_usage()
dataset = sys.argv[1]
base_dir = sys.argv[2]
out_file = sys.argv[3]
names = util.load_names(dataset)
centers = []
for idx, name in enumerate(names):
if dataset == 'nyu': # use synthetic image to compute center
name = name.replace('depth', 'synthdepth')
img = util.load_image(dataset, os.path.join(base_dir, name))
if dataset == 'icvl':
center = util.get_center(img, upper=500, lower=0)
elif dataset == 'nyu':
center = util.get_center(img, upper=1300, lower=500)
elif dataset == 'msra':
center = util.get_center(img, upper=1000, lower=10)
centers.append(center.reshape((1, 3)))
if idx % 500 == 0:
print('{}/{}'.format(idx + 1, len(names)))
util.save_results(centers, out_file)
########################
## This is the preprocessing and validation script for Depth.
## Keep the segment you would like to use and comment out the rest
## this is because there are multiple data files (labels before/after normalization) that will cause conflicts
########################
dataset = 'fpad'
phase = 'train' ## test/train
root_dir = '/home/bilbeisi/REN/'
############################ Draw pose on depth samples #################################
### draw pose on some depth samples to validate world2pixel and image/label loading
### this segment is only for validation
##############################################################################################
lbls = util.load_labels(dataset, phase) ### load test/train data
names = util.load_names(dataset, phase)
centers = util.load_centers(dataset, phase).astype(float)
for idx, name in enumerate(names):
if idx % 1000 == 0:
lbl = np.asarray(np.reshape(lbls[idx].split(), (21, 3)),
dtype=np.float32)
lbl, skel_camcoords = util.world2pixel(lbl, dataset)
img = util.load_image(dataset, os.path.join(root_dir, name))
img /= 1160
img *= 255
points = centers[idx]
img = util.draw_pose(dataset, img, lbl, 3, (255, 0, 0), points)
cv2.imwrite(
root_dir + 'samples/depth/' + phase + '_' + str(idx) + '.png', img)
##############################################################################################
# Preprocess images before running this script
# Remember to change dataset depending on type of data (fpad for depth, and fpac for rgb)
# Run for both phase='test' and phase='train' to get both datafiles
##################################
################ Confirm dataset before running this script! ###############################
dataset = 'fpad' # fpac or fpad
DIR = "../labels/"
phase = 'test' # test/train
h5_fn = os.path.join(DIR, (dataset + '_' + phase + '_data.h5'))
############## Dir containing preprocessed images ###############################
base_dir = '/home/bilbeisi/REN/cropped/'
names = util.load_names(dataset, phase)
labels = util.load_labels(dataset, phase)
if dataset == 'fpad':
imgs = np.zeros((len(names), 1, 96, 96), dtype=np.float32) # depth
else:
imgs = np.zeros((len(names), 3, 96, 96), dtype=np.float32)
lbls = np.zeros((len(labels), 63), dtype=np.float)
for idx, name in enumerate(names):
if dataset == 'fpac':
name = name.replace('.jpeg', '.png')
img = util.load_image(dataset, os.path.join(base_dir, name))
img = img.astype(float)
# revert back to normalized -1,1 since images where saved in 0,255 to allow viewing/verifying
img[:] *= 2
import time
import image_loops
##################################
# This script creates hdf5 files for RGB-D data since different processing is required
# Run for both phase='test' and phase='train' to get both datafiles
# colormap and depth2cords cython functions are used here
##################################
DIR = "../labels/"
phase = 'train'
h5_fn = os.path.join(DIR, ('rgbd_' + phase + '_data_.h5'))
base_dir = '/home/bilbeisi/REN/cropped/'
names = util.load_names('fpad', phase)
labels = util.load_labels('fpad', phase)
cnames = util.load_names('fpac', phase)
centers = util.load_centers('fpad', phase).astype(float)
imgs = np.zeros((len(names), 4, 96, 96), dtype=np.float32)
lbls = np.zeros((len(labels), 63), dtype=np.float)
cube_size = 150 # cube size in mm for cropping
for idx, name in enumerate(names):
cname = cnames[idx]
img = util.load_image('fpad', os.path.join('/home/bilbeisi/REN/', name))
img[img == 0] = 1
cimg = util.load_image('fpac', os.path.join('/home/bilbeisi/REN/', cname))
def main():
if len(sys.argv) < 3:
print_usage()
dataset = sys.argv[1]
out_file = sys.argv[2]
# data_dir_dict = {'nyu': config.nyu_data_dir,
# 'icvl': config.icvl_data_dir + "test/Depth",
# 'msra': config.msra_data_dir}
# base_dir = data_dir_dict[dataset] #sys.argv[3]
# name = os.path.join(base_dir, names[0])
batch_size = 64
if len(sys.argv) == 4:
batch_size = int(sys.argv[3])
# generate deploy prototxt
make_baseline_net(os.path.join(ROOT_DIR, '../models'), dataset)
make_pose_ren_net(os.path.join(ROOT_DIR, '../models'), dataset)
names = util.load_names(dataset)
# centers = util.load_centers(dataset)
centers = None
fx, fy, ux, uy = 587.270, 587.270, 326.548, 230.419
# fx, fy, ux, uy = util.get_param(dataset)
lower_ = 1
upper_ = 650
hand_model = ModelPoseREN(
dataset,
lambda img: get_center(img, lower=lower_, upper=upper_),
param=(fx, fy, ux, uy),
use_gpu=True)
if dataset == 'msra':
hand_model.reset_model(dataset, test_id=0)
base_dir = "/media/reborn/Others/Study/Reborn/Github/Pose-REN/test"
# depthName = os.path.join(base_dir, "000000_depth.bin")
depthName = os.path.join(base_dir, "0.img")
imgName = os.path.join(base_dir, "image_0000.png")
if (dataset == "msra"):
# with open(depthName,"rb") as file:
# data = np.fromfile(file, dtype=np.uint32)
# width, height, left, top, right , bottom = data[:6]
# depth = np.zeros((height, width), dtype=np.float32)
# file.seek(4*6)
# data = np.fromfile(file, dtype=np.float32)
# depth[top:bottom, left:right] = np.reshape(data, (bottom-top, right-left))
# depth[depth == 0] = 10000
# print(depth[depth < depth.max()])
# cv2.imwrite("img0.jpg",depth)
with open(depthName, "rb") as file:
data = np.fromfile(file, dtype=np.uint16)
height = 480
width = 640
# depth = np.zeros((height, width), dtype=np.uint16)
depth = np.reshape(data, (height, width)).astype(np.float32)
min = depth.min()
max = depth.max()
print("min = {}, max = {}".format(min, max))
flag = np.logical_xor(depth <= upper_, depth >= lower_)
depth[flag] = 0
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (7, 7))
depth = cv2.morphologyEx(depth, cv2.MORPH_CLOSE, kernel)
depth[depth == 0] = 10000
# depth[depth == 0] = depth.max()
# with open(depthName,"rb") as file:
# depth = []
# cnt = 0
# contents = iter(partial(file.read, 2), b'')
# for r in contents:
# r_int = int.from_bytes(r, byteorder='big') #将 byte转化为 int
# cnt += 1
# depth.append(r_int)
# # print("i = {} -- {}".format(cnt,r_int))
# depth = np.array(depth)
# depth = np.reshape(depth,(480,640))
# depth[depth == 0] = 10000
elif (dataset == "icvl"):
depth = cv2.imread(imgName, 2)
depth[depth == 0] = depth.max() # invalid pixel
depth = depth.astype(float)
# depth = np.reshape(depth,(240,320))
# depth[depth == 0] = depth.max()
print("np.shape(depth) = {}".format(np.shape(depth)))
# depth = depth[:, ::-1]
# print("names = {}".format(imgName))
# print("np.shape(img) = {}".format(np.shape(img)))
# results = hand_model.detect_files(base_dir, names, centers, max_batch=batch_size)
results = hand_model.detect_image(depth)
print("results = {}".format(results))
print(np.shape(results))
img_show = show_results(depth, results, dataset)
cv2.imwrite('result.jpg', img_show)
cv2.waitKey()
