SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.16133 · 2023

OVO: Open-Vocabulary Occupancy

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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 4 have no contract verdict; 5 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/2305.16133.json · badge: /api/ran/2305.16133.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
dzcgaara/OVO canonical (classifier label)
classifier basis: census
9 ran · 2 no recorded run
FunctionStatusWhere it lives
KL_sep Ran
ran; our contract draft was wrong, not the code
dzcgaara/OVO/ovo/loss/ssc_loss.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from PKU-ICST-MIPL/MRA, is pointer only (licence: NONE) · get_code("033da7a511cc1166")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from PKU-ICST-MIPL/MRA, 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.

batch_dice_loss Ran
ran on a synthesized input
dzcgaara/OVO/ovo/loss/matcher.py
code served (permissive licence) · get_code("690e5487ce07509a")
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 with distinct dimensions
# inputs: shape (batch_size=2, num_elements=4) - predictions
inputs = torch.rand(2, 4, dtype=torch.float32)

# targets: shape (batch_size=2, num_elements=4) - binary labels
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_mask_matching_loss Ran
ran on a synthesized input
dzcgaara/OVO/ovo/loss/mask_matching_loss.py
code served (permissive licence) · get_code("3219ddcaadf018ca")
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)

# Analyzing the function:
# - mask_gts: indexed with [i][j], .unsqueeze(0), torch.sum()
#   Shape: (batch, num_gts, height, width)
# - mask_pred: indexed with [i], .shape[1] accessed, used in repeat and get_dice_loss
#   Shape: (batch, num_pred, height, width)
# Both are 4D tensors used in dice loss computation

batch_size = 2
num_gts = 3
num_pred = 4
height = 5
width = 6

# mask_gts: (batch=2, num_gts=3, height=5, width=6)
# Create with some ground truth masks (some empty, some with values)
mask_gts = torch.zeros(batch_size, num_gts, height, width)
mask_gts[0, 0] = torch.rand(height, width) * 0.5  # has values
mask_gts[0, 1] = torch.zeros(height, width)  # empty, will trigger break
mask_gts[1, 0] = torch.rand(height, width) * 0.3  # has values

# mask_pred: (batch=2, num_pred=4, height=5, width=6)
# Create predictions as probabilities
mask_pred = torch.rand(batch_size, num_pred, height, width)

ARGS = {
    "mask_gts": mask_gts,
    "mask_pred": mask_pred,
}
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_super_CP_multilabel_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
dzcgaara/OVO/ovo/loss/CRP_loss.py
code served (permissive licence) · get_code("5530781a69d17464")
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)

# Analyze function usage:
# - pred_logits: indexed as [i, :, :, :], shape (bs, n_relations, ?, ?)
# - CP_mega_matrices: indexed as [i], shape (bs, ?, ?)
# - Both are reshaped and concatenated, then used in BCEWithLogitsLoss

bs = 2
n_relations = 3
spatial_dim_1 = 4
spatial_dim_2 = 5

# pred_logits shape: (bs, n_relations, spatial_dim_1, spatial_dim_2)
pred_logits = torch.randn(bs, n_relations, spatial_dim_1, spatial_dim_2)

# CP_mega_matrices: list of tensors, one per batch
# Each should have shape (n_relations, spatial_dim_1, spatial_dim_2) to match reshaping logic
CP_mega_matrices = [
    torch.randint(0, 2, (n_relations, spatial_dim_1, spatial_dim_2)).float()
    for _ in range(bs)
]

ARGS = {
    "pred_logits": pred_logits,
    "CP_mega_matrices": CP_mega_matrices,
}
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
focal_loss Ran
ran on a synthesized input
dzcgaara/OVO/ovo/loss/mask_matching_loss.py
code served (permissive licence) · get_code("3a8c2b6651437097")
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: predictions (probabilities between 0 and 1)
# targets: binary labels (0 or 1)
inputs = torch.rand(2, 4)  # batch size 2, 4 elements
targets = torch.randint(0, 2, (2, 4)).float()  # binary labels

ARGS = {
    "inputs": inputs,
    "targets": targets,
    "alpha": 0.25,
    "gamma": 2.0,
}
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
geo_scal_loss Ran
ran; our contract draft was wrong, not the code
dzcgaara/OVO/ovo/loss/ssc_loss.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from PKU-ICST-MIPL/MRA, is pointer only (licence: NONE) · get_code("1cf8e8cd991e96bb")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from PKU-ICST-MIPL/MRA, 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_dice_loss Ran
ran on a synthesized input
dzcgaara/OVO/ovo/loss/mask_matching_loss.py
code served (permissive licence) · get_code("9b0cb366167a3ed4")
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)

# The function expects:
# - input: a tensor that will be reshaped to (batch_size, -1)
# - target: a tensor that will be reshaped to (batch_size, -1) and converted to float

# Create small tensors with distinct dimensions
# batch_size=2, and reshape to different feature dimensions for clarity
batch_size = 2
feature_dim = 4

# input: logits or predictions, shape (2, 4)
input_tensor = torch.randn(batch_size, feature_dim)

# target: binary labels, shape (2, 4)
target_tensor = torch.randint(0, 2, (batch_size, feature_dim))

ARGS = {
    "input": input_tensor,
    "target": target_tensor,
}
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
get_iou Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
dzcgaara/OVO/ovo/loss/sscMetrics.py
code served (permissive licence) · get_code("251f4945130ffbc3")
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 numpy as np

# Analyzing the function:
# - iou_sum: numpy array, shape[0] accessed, indexed by idx, used in division
# - cnt_class: numpy array, same shape as iou_sum, indexed by idx, used in division and slicing [1:]
# The function computes IoU per class and mean IoU across classes (excluding class 0)

# Create small numpy arrays with distinct sizes
# _C (number of classes) = 4
num_classes = 4

# iou_sum: accumulated intersection over union values per class
iou_sum = np.array([100.0, 50.0, 75.0, 25.0], dtype=np.float32)

# cnt_class: count of samples per class (must be non-zero for classes 1-3 to compute mean_iou)
cnt_class = np.array([10, 5, 6, 5], dtype=np.float32)

ARGS = {
    "iou_sum": iou_sum,
    "cnt_class": cnt_class,
}
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 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
sem_scal_loss Ran
ran; our contract draft was wrong, not the code
dzcgaara/OVO/ovo/loss/ssc_loss.py
code served (permissive licence) · get_code("d062b8e225df0d31")
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)

# Analyze the function requirements:
# - pred: shape (batch, n_classes, D1, D2, D3) - 5D tensor, softmax applied on dim 1
# - ssc_target: shape (batch, D1, D2, D3) - 4D tensor with class indices, 255 is mask value
# - Operations: indexing, masking, comparing to class indices, sum operations

# Build small tensors with distinct dimensions for clarity
batch_size = 2
n_classes = 4
d1, d2, d3 = 3, 5, 7

# pred: logits before softmax, shape (batch, n_classes, d1, d2, d3)
pred = torch.randn(batch_size, n_classes, d1, d2, d3)

# ssc_target: class indices (0 to n_classes-1) and 255 for invalid voxels
# shape (batch, d1, d2, d3)
ssc_target = torch.randint(0, n_classes, (batch_size, d1, d2, d3))
# Add some 255 values to create mask variation
ssc_target[0, 0, 0, 0] = 255
ssc_target[1, 1, 2, 3] = 255

ARGS = {
    "pred": pred,
    "ssc_target": ssc_target,
}
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
batch_sigmoid_focal_loss Not yet run
unverified (harvested, no recorded run)
dzcgaara/OVO/ovo/loss/matcher.py
code served (permissive licence) · get_code("3dabeacae742104a")
How it ran

no invocation recorded.

get_accuracy Not yet run
unverified (harvested, no recorded run)
dzcgaara/OVO/ovo/loss/sscMetrics.py
code served (permissive licence) · get_code("40d8d0dfd5686afe")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 6 later) and 5 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
KL_sep 6 other papers
2 dated earlier, 4 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
2312.05752 (2023-12)
Camera-based 3D Semantic Scene Completion with Sparse Guidance Network
and 3 more not named here
geo_scal_loss 6 other papers
2 dated earlier, 4 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
2312.05752 (2023-12)
Camera-based 3D Semantic Scene Completion with Sparse Guidance Network
and 3 more not named here
get_accuracy 5 other papers
3 dated earlier, 2 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2208.08307 (2022-08)
SC-Explorer: Incremental 3D Scene Completion for Safe and Efficient Exploration Mapping and Planning
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
and 2 more not named here
get_iou 5 other papers
3 dated earlier, 2 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2208.08307 (2022-08)
SC-Explorer: Incremental 3D Scene Completion for Safe and Efficient Exploration Mapping and Planning
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
and 2 more not named here
sem_scal_loss 4 other papers
3 dated earlier, 1 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
2303.09551 (2023-03)
SurroundOcc: Multi-Camera 3D Occupancy Prediction for Autonomous Driving
and 1 more not named here
compute_super_CP_multilabel_loss 3 other papers
2 dated earlier, 1 later
2112.00726 (2021-12)
MonoScene: Monocular 3D Semantic Scene Completion
2302.13540 (2023-02)
OccDepth: A Depth-Aware Method for 3D Semantic Scene Completion
2403.08885 (2024-03)
SLCF-Net: Sequential LiDAR-Camera Fusion for Semantic Scene Completion using a 3D Recurrent U-Net

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

Semantic occupancy prediction aims to infer dense geometry and semantics of surroundings for an autonomous agent to operate safely in the 3D environment. Existing occupancy prediction methods are almost entirely trained on human-annotated volumetric data. Although of high quality, the generation of such 3D annotations is laborious and costly, restricting them to a few specific object categories in the training dataset. To address this limitation, this paper proposes Open Vocabulary Occupancy (OVO), a novel approach that allows semantic occupancy prediction of arbitrary classes but without the need for 3D annotations during training. Keys to our approach are (1) knowledge distillation from a pre-trained 2D open-vocabulary segmentation model to the 3D occupancy network, and (2) pixel-voxel filtering for high-quality training data generation. The resulting framework is simple, compact, and compatible with most state-of-the-art semantic occupancy prediction models. On NYUv2 and SemanticKITTI datasets, OVO achieves competitive performance compared to supervised semantic occupancy prediction approaches. Furthermore, we conduct extensive analyses and ablation studies to offer insights into the design of the proposed framework. Our code is publicly available at https://github.com/dzcgaara/OVO.

For agents

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

get_harvested_code_for_paper("2305.16133")
get_code_for_paper("2305.16133")
have("2305.16133")

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

curl https://syntology.ai/api/ran/2305.16133.json

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

[![Syntology run record](https://syntology.ai/api/ran/2305.16133.svg)](https://syntology.ai/paper/2305.16133)

Connect an agent — have() is free.