def main(argv, common_opts):
args = parse_args(argv)
seed_all(12345)
init_algorithms(deterministic=True)
torch.set_grad_enabled(False)
device = common_opts['device']
assert args.multicrop == False, 'TODO: Implement multi-crop for single image inference.'
model = load_model(args.model).to(device).eval()
input_specs: ImageSpecs = model.data_specs.input_specs
image: PIL.Image.Image = PIL.Image.open(args.image, 'r')
image.thumbnail((input_specs.width, input_specs.height))
inp = input_specs.convert(image).to(device, torch.float32)
output = model(inp[None, ...])[0]
norm_skel3d = ensure_cartesian(output.to(CPU, torch.float64), d=3)
fig = plt.figure(figsize=(16, 8))
ax1 = fig.add_subplot(1, 2, 1)
ax2: Axes3D = fig.add_subplot(1, 2, 2, projection='3d')
ax1.imshow(input_specs.unconvert(inp.to(CPU)))
plot_skeleton_on_axes3d(norm_skel3d, CanonicalSkeletonDesc, ax2, invert=True)
plt.show()
def main(argv, common_opts):
args = parse_args(argv)
seed_all(12345)
init_algorithms(deterministic=True)
torch.set_grad_enabled(False)
device = common_opts['device']
if args.model:
# model = load_model(args.model).to(device).eval()
model = load_model(args.model).eval()
data_specs = model.data_specs
else:
model = None
data_specs = DataSpecs(
ImageSpecs(224,
mean=ImageSpecs.IMAGENET_MEAN,
stddev=ImageSpecs.IMAGENET_STDDEV),
JointsSpecs(CanonicalSkeletonDesc, n_dims=3),
)
dataset = get_dataset(args.dataset, data_specs, use_aug=False)
app = MainGUIApp(dataset, device, model)
app.mainloop()
def main(argv, common_opts):
args = parse_args(argv)
seed_all(12345)
init_algorithms(deterministic=True)
torch.set_grad_enabled(False)
device = common_opts['device']
model = load_model(args.model).to(device).eval()
dataset = get_dataset(args.dataset, model.data_specs, use_aug=False)
if args.multicrop:
dataset.multicrop = True
loader = make_unbatched_dataloader(dataset)
else:
loader = make_dataloader(dataset, batch_size=1)
if args.dataset.startswith('h36m-'):
known_depth = True
included_joints = list(range(CanonicalSkeletonDesc.n_joints))
else:
known_depth = False
included_joints = [
CanonicalSkeletonDesc.joint_names.index(joint_name)
for joint_name in VNect_Common_Skeleton
]
print('Use ground truth root joint depth? {}'.format(known_depth))
print('Number of joints in evaluation: {}'.format(len(included_joints)))
df = run_evaluation_3d(model,
device,
loader,
included_joints,
known_depth=known_depth,
print_progress=True)
print('### By sequence')
print()
print(
tabulate(df.drop(columns=['activity_id']).groupby('seq_id').mean(),
headers='keys',
tablefmt='pipe'))
print()
print('### By activity')
print()
print(
tabulate(df.drop(columns=['seq_id']).groupby('activity_id').mean(),
headers='keys',
tablefmt='pipe'))
print()
print('### Overall')
print()
print(
tabulate(
df.drop(columns=['activity_id', 'seq_id']).mean().to_frame().T,
headers='keys',
tablefmt='pipe'))
def main():
args = parse_args()
if (args.mode == 'I' or args.mode == 'i'):
filename = os.path.basename(args.path)
filename_noext = os.path.splitext(filename)[0]
model = load_model(args.model).to(CPU).eval()
coords_img, coords_raw, img_input, img_skele3d = infer_joints(model, args.path)
# print(img_skele3d.shape)
img_skele3d = PIL.Image.fromarray(img_skele3d)
# Calculate step angle
vector1 = coords_img[8,:2] - coords_img[10,:2]
vecto2 = (0,-1)
print(angleBetween(vector1, vecto2)*180/np.pi, "right")
vector1 = coords_img[11,:2] - coords_img[13,:2]
# vector1=(.5,0,0)
vecto2 = (0,-1)
print(angleBetween(vector1, vecto2)*180/np.pi, "left")
draw_skele_on_image(img_input, coords_img[:,:2])
# Used for qualitative evaluation
# draw_color_joints(img_input, coords_img[:,:2])
joints_loc = output_to_JSON(coords_raw, filename_noext)
# img_skele3d.show()
img_skele3d.save('./outputs/3d/' + filename_noext + '.png')
img_input.save('./outputs/' + filename)
#with open('./outputs/joint_loc.json', 'w') as fp:
# json.dump(joints_loc, fp, indent=4)
MayaExporter.WriteToMayaAscii('./outputs/3d/' + filename_noext + '.ma', joints_loc)
MayaExporter.WriteAtomFile('./outputs/3d/' + filename_noext + '_anim.atom', 'C:/Users/imagi/Documents/maya/projects/default/scenes/sk_mannikin_margipose.0007.ma', 1, 1, (coords_raw*100))
if(args.mode =='D' or args.mode == 'd'):
files = os.listdir(args.path)
# print(files)
start = time.time()
model = load_model(args.model).to(CPU).eval()
end = time.time()
print(end-start, "To load Model")
count = 0
joints_loc_list = []
for image in files:
start = time.time()
filename_noext = os.path.splitext(image)[0]
coords_img, coords_raw, img_input, img_skele3d = infer_joints(model, args.path+image)
# print(filename_noext)
img_skele3d = PIL.Image.fromarray(img_skele3d)
draw_skele_on_image(img_input, coords_img[:,:2])
# Used for qualitative evaluation
# draw_color_joints(img_input, coords_img[:,:2])
joints_loc = output_to_JSON(coords_raw, filename_noext)
img_skele3d.save('./outputs/3d/' + filename_noext + '.png')
img_input.save('./outputs/' + image)
joints_loc_list.append(joints_loc)
count += 1
print(time.time()-start, "(s)", "completed " + str(count) + "/" + str(len(files)))
MayaExporter.WriteToMayaAscii('./outputs/3d/' + filename_noext + '.ma', joints_loc)
MayaExporter.WriteAtomFile('./outputs/3d/' + filename_noext + '_anim.atom', 'C:/Users/imagi/Documents/maya/projects/default/scenes/sk_mannikin_margipose.0007.ma', 1, 1, (coords_raw*100))
if(args.mode =='V' or args.mode == 'v'):
filename = os.path.basename(args.path)
filename_noext = os.path.splitext(filename)[0]
(frameArray, fps) = VideoFrames.ExtractFrames(args.path)
frameArray = np.asarray(frameArray, dtype=np.uint8)
skel3DArray = np.zeros((480, 640, 3, frameArray.shape[3]), dtype=np.uint8)
finalFrameArray = np.zeros((256, 256, 3, frameArray.shape[3]), dtype=np.uint8)
strideAngles = np.zeros((frameArray.shape[3],3))
# print(frameArray.shape[3])
start = time.time()
model = load_model(args.model).to(CPU).eval()
end = time.time()
print(end-start, "(s) to load Model")
sign = -1
for i in range(frameArray.shape[3]):
# for i in range(10):
start = time.time()
img = PIL.Image.fromarray(frameArray[:,:,:,i][..., ::-1])
coords_img, coords_raw, img_scaled, skel3DArray[:,:,:,i] = infer_joints(model, img)
# Calculate step angle
vector1 = coords_img[8,:2] - coords_img[10,:2]
vecto2 = (0,-1)
rightStride = angleBetween(vector1, vecto2)*180/np.pi
# print(angleBetween(vector1, vecto2)*180/np.pi, "right")
vector1 = coords_img[11,:2] - coords_img[13,:2]
# vector1=(.5,0,0)
vecto2 = (0,-1)
leftStride = angleBetween(vector1, vecto2)*180/np.pi
# print(angleBetween(vector1, vecto2)*180/np.pi, "left")
strideAngles[i,0] = (i+1)/fps
strideAngles[i,1] = rightStride
strideAngles[i,2] = leftStride
draw_skele_on_image(img_scaled, coords_img[:,:2])
# draw_evaluation_image(img_scaled, coords_img[:,:2], rightStride, leftStride)
# img3d = PIL.Image.fromarray(skel3DArray[:,:,:,i])
finalFrameArray[:,:,:,i] = np.array(img_scaled, dtype=np.uint8)
# skel3DArray[:,:,:,i] = np.array(img3d, dtype=np.uint8)
print(time.time()-start, "(s)", "frames completed " + str(i+1) + "/" + str(frameArray.shape[3]))
# img3d.save('./outputs/3d/'+'skel3d_'+str(i)+'.jpg')
# PIL.Image.SAVE(skel3DArray[:,:,:,-1])
# plt.imsave()
# plt.show()
VideoFrames.FrametoVid(finalFrameArray, skel3DArray, fps, filename_noext)
plt.subplot(2,1,1)
plt.plot(strideAngles[:,0], strideAngles[:,1], color='red')
plt.title('Right Leg')
plt.ylim(0,30)
plt.ylabel('Rect Swing Angle (deg)')
plt.xlim(0,6)
plt.xlabel('(s)')
plt.subplot(2,1,2)
plt.plot(strideAngles[:,0], strideAngles[:,2], color='blue')
plt.title('Left Leg')
plt.ylim(0,30)
plt.ylabel('Rect Swing Angle (deg)')
plt.xlim(0,6)
plt.xlabel('(s)')
plt.tight_layout()
plt.show()