SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2502.14616 · 2025

Monocular Depth Estimation and Segmentation for Transparent Object with Iterative Semantic and Geometric Fusion

arXiv · PDF · Open in the Atlas

Code that ran

11 of 13 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, 1 violated it and 10 have no contract verdict; 0 failures of our instrument; 2 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/2502.14616.json · badge: /api/ran/2502.14616.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
l-j-yuan/modest canonical (classifier label)
classifier basis not recorded on the link
11 ran · 2 no recorded run
FunctionStatusWhere it lives
color_img_to_gray Ran
ran on a synthesized input
l-j-yuan/modest/utils/visualize.py
code served (permissive licence) · get_code("e5551f1a55ca490a")
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 ndarray, ndim 3, dtype family float, shape (4, 5, 3) 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 cv2
import numpy as np

# Create a small color image (BGR format)
# Using shape (4, 5, 3) for height=4, width=5, channels=3
image = np.random.randint(0, 256, size=(4, 5, 3), dtype=np.uint8)

ARGS = {
    "image": image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (4, 5, 3) — shape (4, 5, 3) 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
computeMAE Ran
ran on a synthesized input
l-j-yuan/modest/utils/evaluate.py
code served (permissive licence) · get_code("6b9a82d7ec734a28")
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 float, 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 torch

torch.manual_seed(0)

# Create small tensors for the function
batch_size = 2
height = 4
width = 3

# pred: predicted values, shape (batch_size, height, width)
pred = torch.randn(batch_size, height, width)

# gt: ground truth values, shape (batch_size, height, width)
gt = torch.randn(batch_size, height, width)

# mask: binary mask, shape (batch_size, height, width)
# Using long tensor and then converted to float in the function
mask = torch.randint(0, 2, (batch_size, height, width))

ARGS = {
    "pred": pred,
    "gt": gt,
    "mask": mask
}
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 float, 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
computeRMSE Ran
ran on a synthesized input
l-j-yuan/modest/utils/evaluate.py
code served (permissive licence) · get_code("6a9c1f914792dbd2")
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 float, 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 torch

torch.manual_seed(0)

# Create small tensors with distinct dimensions for clarity
# pred: predicted values, shape (batch=2, height=4, width=3)
pred = torch.randn(2, 4, 3)

# gt: ground truth values, same shape as pred
gt = torch.randn(2, 4, 3)

# mask: binary mask indicating valid pixels, same spatial dims as pred
mask = torch.randint(0, 2, (2, 4, 3))

ARGS = {
    "pred": pred,
    "gt": gt,
    "mask": mask,
}
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 float, 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
compute_depth_metrics Ran
ran on a synthesized input
l-j-yuan/modest/utils/evaluate.py
code served (permissive licence) · get_code("3604976b3f6cf9a4")
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 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 torch.nn as nn

torch.manual_seed(0)

# Create small tensors with distinct dimensions for clarity
batch_size = 2
height = 4
width = 3

# preds: batch x height x width (normalized depth predictions)
preds = torch.rand(batch_size, height, width) * 0.5 + 0.25

# gts: batch x height x width (normalized ground truth depths)
gts = torch.rand(batch_size, height, width) * 0.6 + 0.2

# depth_min and depth_max for denormalization
depth_min = 0.1
depth_max = 10.0

# zero_masks: batch x height x width (boolean mask indicating valid pixels)
zero_masks = torch.ones(batch_size, height, width, dtype=torch.bool)
zero_masks[0, 0, 0] = False  # Mark some pixels as invalid

# gt_masks: batch x height x width (class masks, values 0 to num_classes-1)
num_classes = 2
gt_masks = torch.randint(0, num_classes, (batch_size, height, width))

# denorm: whether to denormalize predictions and ground truth
denorm = True

ARGS = {
    "preds": preds,
    "gts": gts,
    "depth_min": depth_min,
    "depth_max": depth_max,
    "zero_masks": zero_masks,
    "denorm": denorm,
    "gt_masks": gt_masks,
    "num_classes": num_classes,
}
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, ndim 1, 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
depth_to_xyz Ran
ran on a synthesized input
l-j-yuan/modest/utils/loss.py
code served (permissive licence) · get_code("76a7fa32cea967f6")
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 Tensor, ndim 4, dtype family float, shape (2, 3, 4, 3) 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 torch

torch.manual_seed(0)

# Create a small depth image tensor with shape [B, C, H, W]
# B=2 (batch size), C=1 (single channel for depth), H=4 (height), W=3 (width)
depthImage = torch.rand(2, 1, 4, 3) * 1000  # depth values in mm

# Dataset name - must be one of 'syntodd' or 'clearpose'
dataset_name = 'syntodd'

ARGS = {
    'depthImage': depthImage,
    'dataset_name': dataset_name
}
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 Tensor, ndim 4, dtype family float, shape (2, 3, 4, 3) — shape (2, 3, 4, 3) 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
get_losses Ran
ran on a synthesized input
l-j-yuan/modest/utils/builder.py
code served (permissive licence) · get_code("0e8b2c2d2d81f5d1")
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 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 torch.nn as nn

torch.manual_seed(0)

class MaskedL1Loss(nn.Module):
    def forward(self, a, b):
        return 0

class MaskedSmoothL1Loss(nn.Module):
    def forward(self, a, b):
        return 0

class GradientLoss(nn.Module):
    def forward(self, a, b):
        return 0

class SegIouLoss(nn.Module):
    def __init__(self, num_classes):
        super().__init__()
        self.num_classes = num_classes
    
    def forward(self, a, b):
        return 0

config = {
    'Trainer': {
        'loss_depth': 'mse',
        'loss_segmentation': 'ce'
    },
    'Model': {
        'type': 'depth_seg'
    },
    'Dataset': {
        'dataset_name': 'test_dataset',
        'test_dataset': {
            'classes': ['class1', 'class2', 'class3']
        }
    }
}

ARGS = {
    'config': config
}
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, ndim 1, 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
get_surface_normal Ran
ran on a synthesized input
l-j-yuan/modest/utils/loss.py
code served (permissive licence) · get_code("5defdf8118554429")
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 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 torch

torch.manual_seed(0)

# Create a small depth image tensor
# Shape: (B, C, H, W) where B=2 (batch), C=1 (depth channel), H=4, W=3
depthImage = torch.rand(2, 1, 4, 3) * 1000  # depth values in mm range

# Dataset name - must be one of 'syntodd' or 'clearpose'
dataset_name = 'syntodd'

ARGS = {
    'x': depthImage,
    'dataset_name': dataset_name
}
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, ndim 5, 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
gradient Ran
ran on a synthesized input
l-j-yuan/modest/utils/loss.py
code served (permissive licence) · get_code("65701c6ae029f468")
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 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 torch
import torch.nn.functional as F

torch.manual_seed(0)

# The function takes a single parameter 'x' which should be a 4D tensor
# Based on the function body, it performs padding and slicing operations
# expecting shape (batch, channels, height, width)

x = torch.randn(2, 3, 4, 5)

ARGS = {
    "x": x
}
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, ndim 5, 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
isnan Ran
ran, violated the contract we drafted
l-j-yuan/modest/datasets/syntodd.py
pointer only (licence: MIT); the stored copy get_code serves, from sidak/otfusion, is pointer only (licence: NONE) · get_code("5f4f4355be28c01e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from sidak/otfusion, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

multiview_collate Ran
ran on a synthesized input
l-j-yuan/modest/utils/builder.py
code served (permissive licence) · get_code("3e4b6c7dd7e0cfe6")
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)

# Construct a batch of multiview data
# Each element in the batch is a tuple of:
# (images, camera_poses, intrinsics, item3, item4, item5, item6, item7, item8, item9)

batch_element_1 = (
    torch.randn(3, 4, 4),      # images: 3 channels, 4x4
    torch.randn(4, 4),         # camera_poses: 4x4
    torch.randn(3, 3),         # intrinsics: 3x3
    "metadata_1",              # item3
    42,                        # item4
    3.14,                      # item5
    [1, 2, 3],                 # item6
    "unused_1",                # item7
    "unused_2",                # item8
    {"key": "value_1"}         # item9
)

batch_element_2 = (
    torch.randn(3, 4, 4),      # images: 3 channels, 4x4
    torch.randn(4, 4),         # camera_poses: 4x4
    torch.randn(3, 3),         # intrinsics: 3x3
    "metadata_2",              # item3
    43,                        # item4
    2.71,                      # item5
    [4, 5, 6],                 # item6
    "unused_1",                # item7
    "unused_2",                # item8
    {"key": "value_2"}         # item9
)

batch = [batch_element_1, batch_element_2]

ARGS = {
    "batch": batch
}
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
vis_error_map Ran
ran on a synthesized input
l-j-yuan/modest/utils/visualize.py
code served (permissive licence) · get_code("88bdb332e36a05f3")
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 ndarray, ndim 3, dtype family int, shape (4, 3, 3) 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 torch
import numpy as np

torch.manual_seed(0)

# Create small test tensors for depth prediction and ground truth
# Shape: (4, 3) - small 2D depth maps
depth_pred = torch.randn(4, 3)
depth_gt = torch.randn(4, 3)

# The function accepts:
# - depth_pred: can be torch tensor or numpy array
# - depth_gt: can be torch tensor or numpy array
# - type: string indicating the type of visualization (e.g., 'depth')

ARGS = {
    'depth_pred': depth_pred,
    'depth_gt': depth_gt,
    'type': 'depth'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (4, 3, 3) — shape (4, 3, 3) 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
decompress_datapoint Not yet run
unverified (harvested, no recorded run)
l-j-yuan/modest/datasets/syntodd.py
code served (permissive licence) · get_code("2700fc830014d4b3")
How it ran

no invocation recorded.

vis_depth Not yet run
unverified (harvested, no recorded run)
l-j-yuan/modest/utils/visualize.py
code served (permissive licence) · get_code("c8616f3565a0de3a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 12 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.

FunctionAlso present inEarliest-datedOthers named
isnan 16 other papers
13 dated earlier, 3 later · 2 only as an assembly brick
1505.04597 (2015-05)
U-Net: Convolutional Networks for Biomedical Image Segmentation
1910.05653 (2019-10)
Model Fusion via Optimal Transport
2008.07043 (2020-08)
Oriented Object Detection in Aerial Images with Box Boundary-Aware Vectors
and 13 more not named here

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

Transparent object perception is indispensable for numerous robotic tasks. However, accurately segmenting and estimating the depth of transparent objects remain challenging due to complex optical properties. Existing methods primarily delve into only one task using extra inputs or specialized sensors, neglecting the valuable interactions among tasks and the subsequent refinement process, leading to suboptimal and blurry predictions. To address these issues, we propose a monocular framework, which is the first to excel in both segmentation and depth estimation of transparent objects, with only a single-image input. Specifically, we devise a novel semantic and geometric fusion module, effectively integrating the multi-scale information between tasks. In addition, drawing inspiration from human perception of objects, we further incorporate an iterative strategy, which progressively refines initial features for clearer results. Experiments on two challenging synthetic and real-world datasets demonstrate that our model surpasses state-of-the-art monocular, stereo, and multi-view methods by a large margin of about 38.8%-46.2% with only a single RGB input. Codes and models are publicly available at https://github.com/L-J-Yuan/MODEST.

For agents

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

get_harvested_code_for_paper("2502.14616")
get_code_for_paper("2502.14616")
have("2502.14616")

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

curl https://syntology.ai/api/ran/2502.14616.json

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

[![Syntology run record](https://syntology.ai/api/ran/2502.14616.svg)](https://syntology.ai/paper/2502.14616)

Connect an agent — have() is free.