Spaces:
Runtime error
Runtime error
import os, argparse, importlib | |
import torch | |
import time | |
import trimesh | |
import numpy as np | |
from MeshAnything.models.meshanything import MeshAnything | |
import datetime | |
from accelerate import Accelerator | |
from accelerate.utils import set_seed | |
from accelerate.utils import DistributedDataParallelKwargs | |
from safetensors import safe_open | |
from mesh_to_pc import process_mesh_to_pc | |
from huggingface_hub import hf_hub_download | |
class Dataset: | |
def __init__(self, input_type, input_list, mc=False): | |
super().__init__() | |
self.data = [] | |
if input_type == 'pc_normal': | |
for input_path in input_list: | |
# load npy | |
cur_data = np.load(input_path) | |
# sample 4096 | |
assert cur_data.shape[0] >= 4096, "input pc_normal should have at least 4096 points" | |
idx = np.random.choice(cur_data.shape[0], 4096, replace=False) | |
cur_data = cur_data[idx] | |
self.data.append({'pc_normal': cur_data, 'uid': input_path.split('/')[-1].split('.')[0]}) | |
elif input_type == 'mesh': | |
mesh_list = [] | |
for input_path in input_list: | |
# load ply | |
cur_data = trimesh.load(input_path) | |
mesh_list.append(cur_data) | |
if mc: | |
print("First Marching Cubes and then sample point cloud, need several minutes...") | |
pc_list, _ = process_mesh_to_pc(mesh_list, marching_cubes=mc) | |
for input_path, cur_data in zip(input_list, pc_list): | |
self.data.append({'pc_normal': cur_data, 'uid': input_path.split('/')[-1].split('.')[0]}) | |
print(f"dataset total data samples: {len(self.data)}") | |
def __len__(self): | |
return len(self.data) | |
def __getitem__(self, idx): | |
data_dict = {} | |
data_dict['pc_normal'] = self.data[idx]['pc_normal'] | |
# normalize pc coor | |
pc_coor = data_dict['pc_normal'][:, :3] | |
normals = data_dict['pc_normal'][:, 3:] | |
bounds = np.array([pc_coor.min(axis=0), pc_coor.max(axis=0)]) | |
pc_coor = pc_coor - (bounds[0] + bounds[1])[None, :] / 2 | |
pc_coor = pc_coor / np.abs(pc_coor).max() * 0.9995 | |
assert (np.linalg.norm(normals, axis=-1) > 0.99).all(), "normals should be unit vectors, something wrong" | |
data_dict['pc_normal'] = np.concatenate([pc_coor, normals], axis=-1, dtype=np.float16) | |
data_dict['uid'] = self.data[idx]['uid'] | |
return data_dict | |
def get_args(): | |
parser = argparse.ArgumentParser("MeshAnything", add_help=False) | |
parser.add_argument('--llm', default="facebook/opt-350m", type=str) | |
parser.add_argument('--input_dir', default=None, type=str) | |
parser.add_argument('--input_path', default=None, type=str) | |
parser.add_argument('--out_dir', default="inference_out", type=str) | |
parser.add_argument('--pretrained_weights', default="MeshAnything_350m.pth", type=str) | |
parser.add_argument( | |
'--input_type', | |
choices=['mesh','pc_normal'], | |
default='pc', | |
help="Type of the asset to process (default: pc)" | |
) | |
parser.add_argument("--codebook_size", default=8192, type=int) | |
parser.add_argument("--codebook_dim", default=1024, type=int) | |
parser.add_argument("--n_max_triangles", default=800, type=int) | |
parser.add_argument("--batchsize_per_gpu", default=1, type=int) | |
parser.add_argument("--seed", default=0, type=int) | |
parser.add_argument("--mc", default=False, action="store_true") | |
parser.add_argument("--sampling", default=False, action="store_true") | |
args = parser.parse_args() | |
return args | |
def load_model(args): | |
model = MeshAnything(args) | |
print("load model over!!!") | |
ckpt_path = hf_hub_download( | |
repo_id="Yiwen-ntu/MeshAnything", | |
filename="MeshAnything_350m.pth", | |
) | |
tensors = {} | |
with safe_open(ckpt_path, framework="pt", device=0) as f: | |
for k in f.keys(): | |
tensors[k] = f.get_tensor(k) | |
model.load_state_dict(tensors, strict=True) | |
print("load weights over!!!") | |
return model | |
if __name__ == "__main__": | |
args = get_args() | |
cur_time = datetime.datetime.now().strftime("%d_%H-%M-%S") | |
checkpoint_dir = os.path.join(args.out_dir, cur_time) | |
os.makedirs(checkpoint_dir, exist_ok=True) | |
kwargs = DistributedDataParallelKwargs(find_unused_parameters=True) | |
accelerator = Accelerator( | |
mixed_precision="fp16", | |
project_dir=checkpoint_dir, | |
kwargs_handlers=[kwargs] | |
) | |
model = load_model(args) | |
# create dataset | |
if args.input_dir is not None: | |
input_list = sorted(os.listdir(args.input_dir)) | |
# only ply, obj or npy | |
if args.input_type == 'pc_normal': | |
input_list = [os.path.join(args.input_dir, x) for x in input_list if x.endswith('.npy')] | |
else: | |
input_list = [os.path.join(args.input_dir, x) for x in input_list if x.endswith('.ply') or x.endswith('.obj') or x.endswith('.npy')] | |
set_seed(args.seed) | |
dataset = Dataset(args.input_type, input_list, args.mc) | |
elif args.input_path is not None: | |
set_seed(args.seed) | |
dataset = Dataset(args.input_type, [args.input_path], args.mc) | |
else: | |
raise ValueError("input_dir or input_path must be provided.") | |
dataloader = torch.utils.data.DataLoader( | |
dataset, | |
batch_size=args.batchsize_per_gpu, | |
drop_last = False, | |
shuffle = False, | |
) | |
if accelerator.state.num_processes > 1: | |
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model) | |
dataloader, model = accelerator.prepare(dataloader, model) | |
begin_time = time.time() | |
print("Generation Start!!!") | |
with accelerator.autocast(): | |
for curr_iter, batch_data_label in enumerate(dataloader): | |
curr_time = time.time() | |
outputs = model(batch_data_label['pc_normal'], sampling=args.sampling) | |
batch_size = outputs.shape[0] | |
device = outputs.device | |
for batch_id in range(batch_size): | |
recon_mesh = outputs[batch_id] | |
recon_mesh = recon_mesh[~torch.isnan(recon_mesh[:, 0, 0])] # nvalid_face x 3 x 3 | |
vertices = recon_mesh.reshape(-1, 3).cpu() | |
vertices_index = np.arange(len(vertices)) # 0, 1, ..., 3 x face | |
triangles = vertices_index.reshape(-1, 3) | |
scene_mesh = trimesh.Trimesh(vertices=vertices, faces=triangles, force="mesh", | |
merge_primitives=True) | |
scene_mesh.merge_vertices() | |
scene_mesh.update_faces(scene_mesh.unique_faces()) | |
scene_mesh.fix_normals() | |
save_path = os.path.join(checkpoint_dir, f'{batch_data_label["uid"][batch_id]}_gen.obj') | |
num_faces = len(scene_mesh.faces) | |
brown_color = np.array([255, 165, 0, 255], dtype=np.uint8) | |
face_colors = np.tile(brown_color, (num_faces, 1)) | |
scene_mesh.visual.face_colors = face_colors | |
scene_mesh.export(save_path) | |
print(f"{save_path} Over!!") | |
end_time = time.time() | |
print(f"Total time: {end_time - begin_time}") |