SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.00246 · CVPR · 2023

ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution

Khoi Nguyen, Tuan Ngo, Binh-Son Hua

arXiv · PDF · Open in the Atlas

Code that ran

16 of 19 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; 1 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 4 failures of our instrument; 3 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/2303.00246.json · badge: /api/ran/2303.00246.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
VinAIResearch/ISBNet canonical (classifier label)
classifier basis not recorded on the link
16 ran · 3 no recorded run
FunctionStatusWhere it lives
calc_square_dist Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/model_utils.py
code served (permissive licence) · get_code("560832d29481933a")
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 3, dtype family float, shape (2, 3, 5) 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
# a: [bs=2, n=3, c=4]
# b: [bs=2, m=5, c=4]
a = torch.randn(2, 3, 4)
b = torch.randn(2, 5, 4)

ARGS = {
    "a": a,
    "b": b,
    "norm": True,
}
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 3, dtype family float, shape (2, 3, 5) — shape (2, 3, 5) 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_dice_cost Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/matcher.py
code served (permissive licence) · get_code("606e2a02e1bda770")
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 2, dtype family float, shape (2, 2) 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 float tensors for inputs and targets
# inputs: predictions for each example (logits before sigmoid)
# targets: binary classification labels (0 or 1)
# Shape: (batch_size=2, num_classes=4)
inputs = torch.randn(2, 4, dtype=torch.float32)
targets = torch.randint(0, 2, (2, 4), dtype=torch.float32)

ARGS = {
    "inputs": inputs,
    "targets": targets,
}
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 2, dtype family float, shape (2, 2) — shape (2, 2) 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_dice_loss Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/criterion.py
code served (permissive licence) · get_code("878f5aebb430a0cd")
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 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 inputs and targets
# inputs: arbitrary shape float tensor (predictions)
# targets: same shape as inputs, binary labels (0 or 1)
batch_size = 2
num_elements = 4

inputs = torch.rand(batch_size, num_elements, dtype=torch.float32)
targets = torch.randint(0, 2, (batch_size, num_elements), dtype=torch.float32)

# num_boxes: used for normalization in the loss computation
num_boxes = 2

# mask: optional, same shape as inputs/targets when provided
mask = torch.ones(batch_size, num_elements, dtype=torch.float32)

ARGS = {
    "inputs": inputs,
    "targets": targets,
    "num_boxes": num_boxes,
    "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 Tensor, 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_sigmoid_focal_loss Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/criterion.py
code served (permissive licence) · get_code("052d20023e79e3bf")
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 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create small tensors for inputs and targets
# inputs: predictions (logits) - shape (batch_size, num_classes)
# targets: binary labels - same shape as inputs
batch_size = 2
num_classes = 4

inputs = torch.randn(batch_size, num_classes)
targets = torch.randint(0, 2, (batch_size, num_classes)).float()

# num_boxes: integer representing number of boxes
num_boxes = 2

# alpha: weighting factor
alpha = 0.25

# gamma: exponent for modulating factor
gamma = 2.0

# mask: optional mask tensor with same shape as inputs
mask = torch.ones_like(inputs)

ARGS = {
    "inputs": inputs,
    "targets": targets,
    "num_boxes": num_boxes,
    "alpha": alpha,
    "gamma": gamma,
    "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 Tensor, 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
evaluate_offset_mae Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/evaluation/point_wise_eval.py
code served (permissive licence) · get_code("b4cd600209f34611")
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 float64, 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 numpy as np
import logging

# Set up a simple logger
logger = logging.getLogger("test_logger")
logger.setLevel(logging.INFO)
handler = logging.StreamHandler()
logger.addHandler(handler)

# Create small sample data for the function
# pred_list: list of prediction arrays
pred_list = [
    np.random.randn(2, 4).astype(np.float32),  # batch 1: 2 samples, 4 dimensions
    np.random.randn(3, 4).astype(np.float32),  # batch 2: 3 samples, 4 dimensions
]

# gt_list: list of ground truth arrays (same shape as pred_list)
gt_list = [
    np.random.randn(2, 4).astype(np.float32),  # batch 1: 2 samples, 4 dimensions
    np.random.randn(3, 4).astype(np.float32),  # batch 2: 3 samples, 4 dimensions
]

# gt_instance_list: list of instance label arrays (1D for each sample)
gt_instance_list = [
    np.array([1, 2, -100], dtype=np.int32)[:2],  # batch 1: 2 instances
    np.array([1, 2, 3], dtype=np.int32),         # batch 2: 3 instances
]

# ignore_label: value to ignore in calculations
ignore_label = -100

ARGS = {
    "pred_list": pred_list,
    "gt_list": gt_list,
    "gt_instance_list": gt_instance_list,
    "ignore_label": ignore_label,
    "logger": logger,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float 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
evaluate_semantic_acc Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/evaluation/point_wise_eval.py
code served (permissive licence) · get_code("caa556fdf3f1079e")
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 float64, 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 numpy as np
from unittest.mock import Mock

# Create a simple logger mock
logger = Mock()
logger.info = Mock()

# Create sample ground truth and prediction lists
# These should be concatenable numpy arrays
gt_list = [
    np.array([[0, 1, 2, 3], [1, 0, 3, 2]]),  # shape (2, 4)
    np.array([[2, 3, 1, 0], [3, 2, 0, 1]])   # shape (2, 4)
]

pred_list = [
    np.array([[0, 1, 2, 3], [1, 0, 3, 2]]),  # shape (2, 4)
    np.array([[2, 3, 1, 0], [3, 2, 0, 1]])   # shape (2, 4)
]

ARGS = {
    'pred_list': pred_list,
    'gt_list': gt_list,
    'ignore_label': -100,
    'logger': logger
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float 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
get_clones Ran
ran; our contract draft was wrong, not the code
VinAIResearch/ISBNet/isbnet/model/blocks.py
pointer only (licence: BSD-3-Clause); the stored copy get_code serves, from arshadshk/SAINT-pytorch, is pointer only (licence: NONE) · get_code("891b8ebab395921f")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (BSD-3-Clause), but get_code serves the stored copy, first harvested from arshadshk/SAINT-pytorch, 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.

get_iou Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/criterion.py
code served (permissive licence) · get_code("d25a8a78b883dc85")
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 1, dtype family float, shape (2,) 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)

# inputs: logits tensor that will be passed through sigmoid
# Should have shape (batch, features) where sigmoid is applied
# and >= comparison is done, then sum over last dimension
inputs = torch.randn(2, 4)

# targets: binary mask tensor, same shape as inputs after sigmoid
# Used in multiplication with inputs_bool and sum operations
targets = torch.randint(0, 2, (2, 4)).float()

# thresh: threshold for binarization (default 0.5)
thresh = 0.5

ARGS = {
    "inputs": inputs,
    "targets": targets,
    "thresh": thresh,
}
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 1, dtype family float, shape (2,) — shape (2,) 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
ln_norm Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/blocks.py
code served (permissive licence) · get_code("083769d411d192ba")
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 GroupNorm. 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 torch.nn as nn

torch.manual_seed(0)

# The function ln_norm takes a single parameter 'x' which is used as:
# nn.GroupNorm(1, x)
# This means x should be the number of channels/features
# GroupNorm expects an integer for the second argument

x = 4  # number of channels

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 GroupNorm — a GroupNorm output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GroupNorm output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
nms_and_merge Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/model_utils.py
code served (permissive licence) · get_code("60642bb25404f38a")
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 minimal valid arguments for nms_and_merge
# proposals_pred: binary matrix (num_proposals x num_points)
# scores: scores for each proposal
# classes: class label for each proposal
# threshold: IoU threshold for merging

num_proposals = 3
num_points = 4

proposals_pred = torch.randint(0, 2, (num_proposals, num_points)).float()
scores = torch.tensor([0.9, 0.7, 0.5])
classes = torch.tensor([0, 0, 1])
threshold = 0.3

ARGS = {
    "proposals_pred": proposals_pred,
    "scores": scores,
    "classes": classes,
    "threshold": threshold,
}
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
read_mesh_vertices Ran
ran, honoured the contract we drafted
VinAIResearch/ISBNet/isbnet/evaluation/instance_eval_util.py
code served (permissive licence) · get_code("71b8cec4a388f189")
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 2, dtype family float, shape (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 numpy as np
from plyfile import PlyData
import tempfile
import os
import struct

# Create a temporary PLY file with vertex data
def create_test_ply_file():
    temp_file = tempfile.NamedTemporaryFile(mode='wb', suffix='.ply', delete=False)
    
    # Write PLY header
    header = b"""ply
format binary_little_endian 1.0
element vertex 3
property float x
property float y
property float z
end_header
"""
    temp_file.write(header)
    
    # Write 3 vertices with x, y, z coordinates
    vertices_data = [
        (0.0, 1.0, 2.0),
        (3.0, 4.0, 5.0),
        (6.0, 7.0, 8.0),
    ]
    
    for x, y, z in vertices_data:
        temp_file.write(struct.pack('<fff', x, y, z))
    
    temp_file.close()
    return temp_file.name

test_ply_filename = create_test_ply_file()

ARGS = {
    "filename": test_ply_filename
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, plyfile — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (3, 3) — shape (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
set_bn_momentum_default Ran
ran; our contract draft was wrong, not the code
VinAIResearch/ISBNet/isbnet/model/module_utils.py
code served (permissive licence) · get_code("0ad12b1e8408e2f9")
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 function. 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 torch.nn as nn

torch.manual_seed(0)

# The function set_bn_momentum_default takes a bn_momentum parameter
# and returns a function that sets the momentum attribute on BatchNorm modules.
# The bn_momentum parameter should be a float value (typically between 0 and 1).

ARGS = {
    "bn_momentum": 0.1
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
sigmoid_bce_loss Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/matcher.py
code served (permissive licence) · get_code("eaa1dc597251ce1b")
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 2, dtype family float, shape (2, 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 torch.nn.functional as F

torch.manual_seed(0)

# Create small tensors with distinct dimensions
# inputs: (num_querys, N) - shape (2, 4) means 2 queries, 4 dimensions
# targets: (num_inst, N) - shape (3, 4) means 3 instances, 4 dimensions
inputs = torch.randn(2, 4)
targets = torch.randint(0, 2, (3, 4)).float()

ARGS = {
    'inputs': inputs,
    'targets': targets,
}
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 2, dtype family float, shape (2, 3) — shape (2, 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
standard_nms Ran
ran on a synthesized input
VinAIResearch/ISBNet/isbnet/model/model_utils.py
code served (permissive licence) · get_code("ea0eed78ca52d7a8")
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)

# Analyzing the function:
# - proposals_pred: used with .type(scores.dtype), .sum(1), indexed with [get_idxs]
#   Shape: (N, M) where N is number of proposals, M is feature dimension
# - categories: indexed with [i], [ixs[1:]], [get_idxs], compared with ==
#   Shape: (N,) integer tensor
# - scores: used with torch.argsort(descending=True), indexed with [get_idxs]
#   Shape: (N,) float tensor
# - boxes: indexed with [get_idxs]
#   Shape: (N, K) where K is box dimension
# - threshold: float, compared with >

# Create small tensors
num_proposals = 2
feature_dim = 4
box_dim = 3

proposals_pred = torch.randn(num_proposals, feature_dim)
categories = torch.tensor([0, 0], dtype=torch.long)
scores = torch.tensor([0.9, 0.5], dtype=torch.float32)
boxes = torch.randn(num_proposals, box_dim)
threshold = 0.2

ARGS = {
    "proposals_pred": proposals_pred,
    "categories": categories,
    "scores": scores,
    "boxes": boxes,
    "threshold": threshold,
}
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
transform_points Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
VinAIResearch/ISBNet/isbnet/evaluation/instance_eval_util.py
code served (permissive licence) · get_code("d328ddb2e66ec4be")
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 2, dtype family float, shape (2, 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 numpy as np

# Construct a 4x4 homography/transformation matrix
# This could represent a 3D affine transformation (perspective or otherwise)
matrix = np.array([
    [1.0, 0.0, 0.0, 0.1],
    [0.0, 1.0, 0.0, 0.2],
    [0.0, 0.0, 1.0, 0.3],
    [0.0, 0.0, 0.0, 1.0]
], dtype=np.float32)

# Construct points array with shape (2, 3) - 2 points, 3 coordinates each
points = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0]
], dtype=np.float32)

ARGS = {
    "matrix": matrix,
    "points": points
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 3) — shape (2, 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
unique_with_inds Ran
ran; our contract draft was wrong, not the code
VinAIResearch/ISBNet/isbnet/model/blocks.py
pointer only (licence: BSD-3-Clause); the stored copy get_code serves, from zhiyuanYU134/MIRETR, is pointer only (licence: NONE) · get_code("6ec42f54dda661e2")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (BSD-3-Clause), but get_code serves the stored copy, first harvested from zhiyuanYU134/MIRETR, 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.

build_matcher Not yet run
unverified (harvested, no recorded run)
VinAIResearch/ISBNet/isbnet/model/matcher.py
code served (permissive licence) · get_code("f117960158fe65f5")
How it ran

no invocation recorded.

evaluate_semantic_miou Not yet run
unverified (harvested, no recorded run)
VinAIResearch/ISBNet/isbnet/evaluation/point_wise_eval.py
code served (permissive licence) · get_code("bc27be3c2a69fe70")
How it ran

no invocation recorded.

load_ids Not yet run
unverified (harvested, no recorded run)
VinAIResearch/ISBNet/isbnet/evaluation/instance_eval_util.py
code served (permissive licence) · get_code("1569c8cb19910d35")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 10 are also present in other papers whose code we hold (10 with at least one paper dated earlier, 1 the same month, 7 later) and 9 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
get_clones 40 other papers
14 dated earlier, 2 the same month, 23 later, 1 same year, order not determinable · 8 only as an assembly brick
1905.12794 (2019-05)
Fashion IQ: A New Dataset Towards Retrieving Images by Natural Language Feedback
2002.07033 (2020-02)
Towards an Appropriate Query, Key, and Value Computation for Knowledge Tracing
2010.12042 (2020-10)
SAINT+: Integrating Temporal Features for EdNet Correctness Prediction
and 37 more not named here
set_bn_momentum_default 16 other papers
6 dated earlier, 10 later
1904.09664 (2019-04)
Deep Hough Voting for 3D Object Detection in Point Clouds
1911.11236 (2019-11)
RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds
2004.05679 (2020-04)
MLCVNet: Multi-Level Context VoteNet for 3D Object Detection
and 13 more not named here
load_ids 5 other papers
3 dated earlier, 1 later, 1 same year, order not determinable
2004.01658 (2020-04)
PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset
2203.16832 (2022-03)
Point Scene Understanding via Disentangled Instance Mesh Reconstruction
and 2 more not named here
read_mesh_vertices 5 other papers
2 dated earlier, 2 later, 1 same year, order not determinable
2203.05238 (2022-03)
Back to Reality: Weakly-supervised 3D Object Detection with Shape-guided Label Enhancement
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset
aaai_25335 (2023 (year only))
Superpoint Transformer for 3D Scene Instance Segmentation
and 2 more not named here
transform_points 3 other papers
1 dated earlier, 1 later, 1 same year, order not determinable
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset
aaai_25335 (2023 (year only))
Superpoint Transformer for 3D Scene Instance Segmentation
2406.02548 (2024-06)
Open-YOLO 3D: Towards Fast and Accurate Open-Vocabulary 3D Instance Segmentation
build_matcher 2 other papers
1 dated earlier, 1 later
2103.12115 (2021-03)
End-to-End Trainable Multi-Instance Pose Estimation with Transformers
2411.10293 (2024-11)
RETR: Multi-View Radar Detection Transformer for Indoor Perception
unique_with_inds 2 other papers
1 dated earlier, 1 later
2207.10859 (2022-07)
Geodesic-Former: a Geodesic-Guided Few-shot 3D Point Cloud Instance Segmenter
2404.04557 (2024-04)
Learning Instance-Aware Correspondences for Robust Multi-Instance Point Cloud Registration in Cluttered Scenes
evaluate_offset_mae 1 other paper
1 dated earlier
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset
evaluate_semantic_acc 1 other paper
1 dated earlier
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset
evaluate_semantic_miou 1 other paper
1 dated earlier
2203.09065 (2022-03)
STPLS3D: A Large-Scale Synthetic and Real Aerial Photogrammetry 3D Point Cloud Dataset

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

Existing 3D instance segmentation methods are predominated by the bottom-up design -manually fine-tuned algorithm to group points into clusters followed by a refinement network. However, by relying on the quality of the clusters, these methods generate susceptible results when (1) nearby objects with the same semantic class are packed together, or (2) large objects with loosely connected regions. To address these limitations, we introduce ISBNet, a novel cluster-free method that represents instances as kernels and decodes instance masks via dynamic convolution. To efficiently generate high-recall and discriminative kernels, we propose a simple strategy named Instance-aware Farthest Point Sampling to sample candidates and leverage the local aggregation layer inspired by PointNet++ to encode candidate features. Moreover, we show that predicting and leveraging the 3D axis-aligned bounding boxes in the dynamic convolution further boosts performance. Our method set new state-of-the-art results on ScanNetV2 (55.9), S3DIS (60.8), and STPLS3D (49.2) in terms of AP and retains fast inference time (237ms per scene on Scan-NetV2). The source code and trained models are available at https://github.com/VinAIResearch/ISBNet.

For agents

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

get_harvested_code_for_paper("2303.00246")
get_code_for_paper("2303.00246")
have("2303.00246")

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

curl https://syntology.ai/api/ran/2303.00246.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.00246.svg)](https://syntology.ai/paper/2303.00246)

Connect an agent — have() is free.