def calculate_mean_and_std():
vnum = SET.VERTS_NUM
case_folder_list = os.listdir(SET.CASE)
md = [np.load(os.path.join(SET.JOINTS_ANIM_PATH))]
case_list = [1]
with open(os.path.join(SET.INFO_BASIS, SET.TRAIN_LIST)) as f:
content = f.readlines()
flist_train = [x.strip().split(' ') for x in content]
random.shuffle(flist_train)
uid = 4000
verts = np.zeros((uid, vnum, 3))
for fid in tqdm(range(uid)):
fr = flist_train[fid]
frid = int(fr[2])
case_id = int(fr[0][6:])
case_index = case_list.index(case_id)
root_joint = 0
myr = np.asarray(md[case_index][frid, root_joint, :9]).reshape((3, 3))
myt = np.asarray(md[case_index][frid, root_joint, 9:]).reshape((1, 3))
npy_path = os.path.join(SET.CASE, case_folder_list[case_id - 1],
fr[1] + '_npy',
str(frid).zfill(5) + '.npy')
x = np.load(npy_path)
y = np.dot(x - np.tile(myt, [vnum, 1]), np.linalg.inv(myr))
verts[fid, :, :] = y.reshape(1, vnum, 3)
for i in range(40):
xid = random.randint(0, uid)
obj.objexport(verts[xid, :, :], SET.MESH_FACES,
os.path.join(SET.DATA_SAMPLE, "s_%s.obj" % str(i)))
m = np.mean(verts, axis=0)
s = np.std(verts, axis=0)
np.save(os.path.join(SET.INFO_BASIS, SET.DATA_MEAN), m)
np.save(os.path.join(SET.INFO_BASIS, SET.DATA_STD), s)
def visualize_skeleton():
md = np.load(SET.JOINTS_ANIM_PATH)
xid = 10
ts = [-1, 1]
fs = [[1, 2, 3], [3, 2, 4], [7, 6, 5], [6, 7, 8], [1, 3, 5], [5, 3, 7],
[2, 6, 4], [4, 6, 8], [3, 4, 7], [7, 4, 8], [1, 5, 2], [2, 5, 6]]
for tid in range(md.shape[1]):
myt = np.asarray(md[xid, tid, 9:]).reshape((1, 3))
vs = [[0, 0, 0]] * 8
ct = -1
for p1 in ts:
for p2 in ts:
for p3 in ts:
ct = ct + 1
vs[ct] = [myt[0, 0] + p1, myt[0, 1] + p2, myt[0, 2] + p3]
file_name = "fr_%s.sk_%s.obj" % (str(xid).zfill(5), str(tid).zfill(5))
obj.objexport(vs, fs, os.path.join(SET.SKELETON, file_name))
z = torch.div(basis_net.g(y) - basis_mean, basis_std)
res = z.clone()
res[z != z] = 0.0
z = res
temp_data = model.g(z, w)
data[bid] = temp_data
zs[bid] = z
weights[bid] = w
loss_1 = loss_1 + torch.sum(data.std(dim=0))
dmean = data.mean(dim=0)
for bid in range(batchsize_2):
q = model.h(dmean, weights[bid])
w = basis_net.h(torch.mul(q, basis_std) + basis_mean)
u_ = np.multiply(w.cpu().data.numpy().reshape(vnum, 3),
data_std) + data_mean
u = np.dot(u_, (myr)) + np.tile(myt, [vnum, 1])
loss_2 = tools.calculate_dist(q, zs[bid])
file_name_grt = 'eval_%s_%s_%s_%s_grt.obj' % (str(bid).zfill(3), fr[0],
fr[1], fr[2])
file_name_rec = 'eval_%s_%s_%s_%s_%s_rec.obj' % (
str(bid).zfill(3), fr[0], fr[1], fr[2], str(float(loss_2.cpu())))
obj.objexport(x, tmp_f, os.path.join(SET.EVAL_MIE, file_name_grt))
obj.objexport(u, tmp_f, os.path.join(SET.EVAL_MIE, file_name_rec))
verts = np.zeros((uid, vnum, 3))
for fid in tqdm(range(uid)):
fr = flist_train[fid]
frid = int(fr[2])
case_id = int(fr[0][6:])
case_index = case_list.index(case_id)
root_joint = 0
myr = np.asarray(md[case_index][frid, root_joint, :9]).reshape((3, 3))
myt = np.asarray(md[case_index][frid, root_joint, 9:]).reshape((1, 3))
npy_path = os.path.join(SET.CASE, case_folder_list[case_id - 1],
fr[1] + '_npy',
str(frid).zfill(5) + '.npy')
x = np.load(npy_path)
y = np.dot(x - np.tile(myt, [vnum, 1]), np.linalg.inv(myr))
verts[fid, :, :] = y.reshape(1, vnum, 3)
for i in range(40):
xid = random.randint(0, uid)
obj.objexport(verts[xid, :, :], SET.MESH_FACES,
os.path.join(SET.DATA_SAMPLE, "s_%s.obj" % str(i)))
m = np.mean(verts, axis=0)
s = np.std(verts, axis=0)
np.save(os.path.join(SET.INFO_BASIS, SET.DATA_MEAN), m)
np.save(os.path.join(SET.INFO_BASIS, SET.DATA_STD), s)
import os from common import obj_operation as obj from common import setting as SET from tqdm import tqdm md = np.load(SET.JOINTS_ANIM_PATH) xid = 10 ts = [-1,1] fs = [[1,2,3],[3,2,4],[7,6,5],[6,7,8],[1,3,5],[5,3,7],[2,6,4],[4,6,8],[3,4,7],[7,4,8],[1,5,2],[2,5,6]] for tid in range(md.shape[1]): myt = np.asarray(md[xid,tid,9:]).reshape((1,3)) vs = [[0,0,0]] *8 ct = -1 for p1 in ts: for p2 in ts: for p3 in ts: ct = ct + 1 vs[ct] = [myt[0,0]+p1, myt[0,1]+p2, myt[0,2]+p3] file_name = "fr_%s.sk_%s.obj" % (str(xid).zfill(5), str(tid).zfill(5)) obj.objexport(vs, fs, os.path.join(SET.SKELETON, file_name))