SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.12494 · 2020

Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency

arXiv · PDF · Open in the Atlas

Code that ran

11 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 0 failures of our instrument; 5 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2007.12494.json · badge: /api/ran/2007.12494.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
jiaxiangshang/MGCNet canonical (classifier label)
classifier basis: census
11 ran · 5 no recorded run
FunctionStatusWhere it lives
compute_3dmm_regular_l2_loss Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_tfGraph/deep_3dmm_loss.py
code served (permissive licence) · get_code("0dd785bfccfeb203")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned EagerTensor, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tensorflow as tf
import numpy as np

# The function expects pred_batch_list which is a list of tensors
# Each tensor will have tf.reduce_sum(tf.square(...)) applied to it
# So we need a list of TensorFlow tensors

# Create small sample tensors for the batch list
pred_batch_1 = tf.constant([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)
pred_batch_2 = tf.constant([[0.5, 1.5, 2.5, 3.5], [4.5, 5.5, 6.5, 7.5]], dtype=tf.float32)
pred_batch_3 = tf.constant([[-1.0, -2.0, -3.0, -4.0], [-5.0, -6.0, -7.0, -8.0]], dtype=tf.float32)

pred_batch_list = [pred_batch_1, pred_batch_2, pred_batch_3]

ARGS = {
    "pred_batch_list": pred_batch_list
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded SymbolicTensor ndim 0, which disagrees with the fixture run
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_lm_eul_square_loss Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_tfGraph/deep_3dmm_loss.py
code served (permissive licence) · get_code("72f280c05557dd1d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned EagerTensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tensorflow as tf

# Create small sample batches
# pred_batch_list: list of predicted landmarks
# Each element should be a tensor of shape (batch_size, num_landmarks, 2)
batch_size = 2
num_landmarks = 4
landmark_dim = 2

# Create prediction batch list with 2 batches
pred_batch_1 = tf.constant([
    [[1.0, 2.0], [3.0, 4.0], [5.0, 6.0], [7.0, 8.0]],
    [[2.0, 3.0], [4.0, 5.0], [6.0, 7.0], [8.0, 9.0]]
], dtype=tf.float32)

pred_batch_2 = tf.constant([
    [[1.5, 2.5], [3.5, 4.5], [5.5, 6.5], [7.5, 8.5]],
    [[2.5, 3.5], [4.5, 5.5], [6.5, 7.5], [8.5, 9.5]]
], dtype=tf.float32)

pred_batch_list = [pred_batch_1, pred_batch_2]

# Create ground truth batch list with same structure
gt_batch_1 = tf.constant([
    [[1.1, 2.1], [3.1, 4.1], [5.1, 6.1], [7.1, 8.1]],
    [[2.1, 3.1], [4.1, 5.1], [6.1, 7.1], [8.1, 9.1]]
], dtype=tf.float32)

gt_batch_2 = tf.constant([
    [[1.6, 2.6], [3.6, 4.6], [5.6, 6.6], [7.6, 8.6]],
    [[2.6, 3.6], [4.6, 5.6], [6.6, 7.6], [8.6, 9.6]]
], dtype=tf.float32)

gt_batch_list = [gt_batch_1, gt_batch_2]

# Create weight tensor of shape (batch_size, num_landmarks)
weight = tf.constant([
    [1.0, 0.5, 1.0, 0.8],
    [0.9, 1.0, 0.7, 1.0]
], dtype=tf.float32)

ARGS = {
    "pred_batch_list": pred_batch_list,
    "gt_batch_list": gt_batch_list,
    "weight": weight
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
decoder_align_depthMap Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_tfGraph/decoder_multiView.py
code served (permissive licence) · get_code("90d49e955236c46c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 5, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tensorflow as tf

class OptConfig:
    def __init__(self, batch_size, img_height, img_width):
        self.batch_size = batch_size
        self.img_height = img_height
        self.img_width = img_width

# Create configuration
opt = OptConfig(batch_size=2, img_height=4, img_width=3)

# Create sample depth maps (batch_size, height, width, 1)
list_depthMap = [
    tf.constant([[[1.0], [2.0], [3.0]], 
                 [[4.0], [5.0], [6.0]], 
                 [[7.0], [8.0], [9.0]], 
                 [[10.0], [11.0], [12.0]]], dtype=tf.float32),
    tf.constant([[[2.0], [3.0], [4.0]], 
                 [[5.0], [6.0], [7.0]], 
                 [[8.0], [9.0], [10.0]], 
                 [[11.0], [12.0], [13.0]]], dtype=tf.float32)
]
# Add batch dimension to match batch_size=2
list_depthMap = [tf.expand_dims(d, 0) for d in list_depthMap]
list_depthMap = [tf.tile(d, [2, 1, 1, 1]) for d in list_depthMap]

# Create synthetic depth maps (same shape)
list_syn_depthMap = [
    tf.constant([[[1.5], [2.5], [3.5]], 
                 [[4.5], [5.5], [6.5]], 
                 [[7.5], [8.5], [9.5]], 
                 [[10.5], [11.5], [12.5]]], dtype=tf.float32),
    tf.constant([[[2.5], [3.5], [4.5]], 
                 [[5.5], [6.5], [7.5]], 
                 [[8.5], [9.5], [10.5]], 
                 [[11.5], [12.5], [13.5]]], dtype=tf.float32)
]
list_syn_depthMap = [tf.expand_dims(d, 0) for d in list_syn_depthMap]
list_syn_depthMap = [tf.tile(d, [2, 1, 1, 1]) for d in list_syn_depthMap]

# Create projection masks (batch_size, height, width, 1) with binary values
list_syn_mask = [
    tf.ones((2, 4, 3, 1), dtype=tf.float32),
    tf.ones((2, 4, 3, 1), dtype=tf.float32)
]

ARGS = {
    'opt': opt,
    'list_depthMap': list_depthMap,
    'list_syn_depthMap': list_syn_depthMap,
    'list_syn_mask': list_syn_mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 5, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
format_file_list Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/face_io.py
code served (permissive licence) · get_code("fe839848957aa1a8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import tempfile
import shutil

# Create a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()
data_root = temp_dir

# Create subdirectories
os.makedirs(os.path.join(data_root, "subfolder1"), exist_ok=True)
os.makedirs(os.path.join(data_root, "subfolder2"), exist_ok=True)

# Create a split file (e.g., "train.txt")
split = "train"
split_file_path = os.path.join(data_root, f"{split}.txt")

# Write sample content to the split file
# Format: "subfolder frame_id\n"
with open(split_file_path, 'w') as f:
    f.write("subfolder1 frame_001\n")
    f.write("subfolder2 frame_002\n")
    f.write("subfolder1 frame_003\n")

# Create dummy image and info files
for subfolder in ["subfolder1", "subfolder2"]:
    for frame_id in ["frame_001", "frame_002", "frame_003"]:
        jpg_path = os.path.join(data_root, subfolder, f"{frame_id}.jpg")
        info_path = os.path.join(data_root, subfolder, f"{frame_id}_info.txt")
        open(jpg_path, 'a').close()
        open(info_path, 'a').close()

ARGS = {
    "data_root": data_root,
    "split": split,
    "fmt": None,
    "sort": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_coeff Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/parse_encoder.py
code served (permissive licence) · get_code("f41ba85e0b565686")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a config object with the required attributes
class OptConfig:
    def __init__(self):
        self.gpmm_rank = 4
        self.gpmm_exp_rank = 3
        
opt = OptConfig()

# Calculate total coefficient dimension
# shape: gpmm_rank (4)
# color: gpmm_rank (4)
# exp: gpmm_exp_rank (3)
# pose_render: 6
# sh: 27
# Total: 4 + 4 + 3 + 6 + 27 = 44

batch_size = 2
total_coeff_dim = 2 * opt.gpmm_rank + opt.gpmm_exp_rank + 6 + 27

# Create coefficient tensor with batch_size x total_coeff_dim
coeff_all = torch.randn(batch_size, total_coeff_dim)

# Create defined_pose_main with shape (batch_size, 6) to match pred_pose_render
defined_pose_main = torch.randn(batch_size, 6)

ARGS = {
    "opt": opt,
    "coeff_all": coeff_all,
    "defined_pose_main": defined_pose_main
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_coeff_list Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/parse_encoder.py
code served (permissive licence) · get_code("66db690af8fb961d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a minimal options object with required attributes
class Opt:
    def __init__(self):
        self.gpmm_rank = 4
        self.gpmm_exp_rank = 3
        # Total expected size: 2*gpmm_rank + gpmm_exp_rank + 6 + 27
        # = 2*4 + 3 + 6 + 27 = 48

opt = Opt()

# Create coeff_all tensor with shape (batch_size, total_coeff_dim)
# Total dim = 2*4 + 3 + 6 + 27 = 48
batch_size = 2
coeff_dim = 2 * opt.gpmm_rank + opt.gpmm_exp_rank + 6 + 27  # 48
coeff_all = torch.randn(batch_size, coeff_dim)

# Create a list of coeff_all tensors for parse_coeff_list
coeff_all_list = [
    torch.randn(batch_size, coeff_dim),
    torch.randn(batch_size, coeff_dim),
]

# Create defined_pose_main tensor with shape (batch_size, 6)
defined_pose_main = torch.randn(batch_size, 6)

ARGS = {
    "opt": opt,
    "coeff_all_list": coeff_all_list,
    "defined_pose_main": defined_pose_main,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_gpu_list Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/format_helper.py
code served (permissive licence) · get_code("c5ff01391d1cf967")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function parse_gpu_list takes a string gpu_list and calls .split(',') on it
# So we need to provide a string that contains comma-separated values

ARGS = {
    'gpu_list': '0,1,2,3'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_self_6DoF Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/face_io.py
code served (permissive licence) · get_code("42d1cd9121f06f56")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import io

# Create a small test file-like object with 6DoF data
test_data = "1.5,2.3,3.1,4.2,5.0,6.7"
f_info = io.StringIO(test_data)

ARGS = {
    "path_info_save": f_info,
    "inter": ","
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_self_lm Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/face_io.py
code served (permissive licence) · get_code("6b85f59c39b1b550")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tempfile
import os

# Create a temporary file with landmark data in the expected format
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("header_line\n")
temp_file.write("1.0,2.0,3.0\n")
temp_file.write("4.5,5.5,6.5\n")
temp_file.write("7.1,8.2,9.3\n")
temp_file.close()

ARGS = {
    "path_info_save": temp_file.name
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_seq Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/format_helper.py
code served (permissive licence) · get_code("1f951817f8f76d19")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function parse_seq takes a list and slices it
# list_seq[0:1] gets the first element as a list
# list_seq[1:] gets all elements from index 1 onward

# Create a simple list with distinct elements to demonstrate the slicing
list_seq = [
    torch.tensor([1, 2, 3, 4]),
    torch.tensor([5, 6, 7, 8]),
    torch.tensor([9, 10, 11, 12]),
]

ARGS = {
    "list_seq": list_seq
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
unpack_image_sequence Ran
ran on a synthesized input
jiaxiangshang/MGCNet/src_common/common/tf_io.py
code served (permissive licence) · get_code("4cb9a56818962f9f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tensorflow as tf

# Create a small image sequence tensor
# The function expects to slice this tensor into target and source images
# Based on the slicing logic:
# - tgt_start_idx = img_width * (num_source // 2)
# - src_image_1: [0, tgt_start_idx)
# - tgt_image: [tgt_start_idx, tgt_start_idx + img_width)
# - src_image_2: [tgt_start_idx + img_width, end)
# So total width needed: tgt_start_idx + img_width + img_width * (num_source // 2)
#                      = img_width * (num_source // 2) + img_width + img_width * (num_source // 2)
#                      = img_width * (num_source + 1)

img_height = 4
img_width = 3
num_source = 2
channel = 2

# Total width needed: img_width * (num_source + 1) = 3 * 3 = 9
total_width = img_width * (num_source + 1)

# Create image sequence tensor with shape [img_height, total_width, channel]
image_seq = tf.constant(
    [[[i + j + k for k in range(channel)] 
      for j in range(total_width)] 
     for i in range(img_height)], 
    dtype=tf.float32
)

ARGS = {
    "image_seq": image_seq,
    "img_height": img_height,
    "img_width": img_width,
    "num_source": num_source
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
data_augmentation_mul Not yet run
unverified (harvested, no recorded run)
jiaxiangshang/MGCNet/src_common/common/tf_io.py
code served (permissive licence) · get_code("7bdf0eba01b8077c")
How it ran

no invocation recorded.

draw_landmark_image Not yet run
unverified (harvested, no recorded run)
jiaxiangshang/MGCNet/src_common/common/visual_helper.py
code served (permissive licence) · get_code("451093c79624fdba")
How it ran

no invocation recorded.

render_lm2d Not yet run
unverified (harvested, no recorded run)
jiaxiangshang/MGCNet/src_common/common/visual_helper.py
code served (permissive licence) · get_code("36c67fa3a4051ff4")
How it ran

no invocation recorded.

render_lm2d_image Not yet run
unverified (harvested, no recorded run)
jiaxiangshang/MGCNet/src_common/common/visual_helper.py
code served (permissive licence) · get_code("874dd70da2cb0d8b")
How it ran

no invocation recorded.

unpack_image_batch_list Not yet run
unverified (harvested, no recorded run)
jiaxiangshang/MGCNet/src_common/common/tf_io.py
code served (permissive licence) · get_code("f041f1cfef34ca7f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recent learning-based approaches, in which models are trained by single-view images have shown promising results for monocular 3D face reconstruction, but they suffer from the ill-posed face pose and depth ambiguity issue. In contrast to previous works that only enforce 2D feature constraints, we propose a self-supervised training architecture by leveraging the multi-view geometry consistency, which provides reliable constraints on face pose and depth estimation. We first propose an occlusion-aware view synthesis method to apply multi-view geometry consistency to self-supervised learning. Then we design three novel loss functions for multi-view consistency, including the pixel consistency loss, the depth consistency loss, and the facial landmark-based epipolar loss. Our method is accurate and robust, especially under large variations of expressions, poses, and illumination conditions. Comprehensive experiments on the face alignment and 3D face reconstruction benchmarks have demonstrated superiority over state-of-the-art methods. Our code and model are released in https://github.com/jiaxiangshang/MGCNet.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2007.12494")
get_code_for_paper("2007.12494")
have("2007.12494")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2007.12494.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2007.12494.svg)](https://syntology.ai/paper/2007.12494)

Connect an agent — have() is free.