SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2004.03572 · 2020

Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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 0 have no contract verdict; 3 failures of our instrument; 0 no recorded run), measured 2026-11-21; 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/2004.03572.json · badge: /api/ran/2004.03572.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
zju3dv/disprcnn canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
filter_bbox_3d Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zju3dv/disprcnn/disprcnn/modeling/pointnet_module/point_rcnn/lib/net/point_rcnn.py
code served (permissive licence) · get_code("e12a7703159f2ff5")
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)

# bbox_3d should be a tensor with 8 vertices of a 3D bounding box
# Each vertex is a 3D point, so shape should be (8, 3)
bbox_3d = torch.tensor([
    [0.0, 0.0, 0.0],  # vertex 0
    [1.0, 0.0, 0.0],  # vertex 1
    [1.0, 1.0, 0.0],  # vertex 2
    [0.0, 1.0, 0.0],  # vertex 3
    [0.0, 0.0, 1.0],  # vertex 4
    [1.0, 0.0, 1.0],  # vertex 5
    [1.0, 1.0, 1.0],  # vertex 6
    [0.0, 1.0, 1.0],  # vertex 7
], dtype=torch.float32)

# point should be a batch of 3D points, shape (batch_size, 3)
# Using batch size 2 with 3D coordinates
point = torch.tensor([
    [0.5, 0.5, 0.5],   # point inside bbox
    [2.0, 2.0, 2.0],   # point outside bbox
], dtype=torch.float32)

ARGS = {
    'bbox_3d': bbox_3d,
    'point': point,
}
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
remove_empty_proposals Ran
ran; our contract draft was wrong, not the code
zju3dv/disprcnn/disprcnn/modeling/pointnet_module/point_rcnn/lib/net/point_rcnn.py
code served (permissive licence) · get_code("3b1a2015d6936cc4")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class MockDetectionResult:
    """Mock object that behaves like a detection result with bbox and indexing support."""
    def __init__(self, bbox):
        self.bbox = bbox
    
    def __getitem__(self, mask):
        """Support boolean indexing to filter results."""
        new_bbox = self.bbox[mask]
        return MockDetectionResult(new_bbox)

# Create sample bounding boxes: [x1, y1, x2, y2]
# Batch size: 2, number of proposals: 4 per batch
left_bbox_1 = torch.tensor([
    [10.0, 20.0, 15.0, 25.0],  # valid (width=5, height=5)
    [30.0, 40.0, 31.0, 41.0],  # valid (width=1, height=1) - edge case
    [50.0, 60.0, 50.5, 60.5],  # invalid (width=0.5, height=0.5)
    [100.0, 100.0, 102.0, 102.0]  # valid (width=2, height=2)
], dtype=torch.float32)

left_bbox_2 = torch.tensor([
    [5.0, 5.0, 10.0, 10.0],  # valid
    [20.0, 20.0, 20.5, 20.5],  # invalid
    [70.0, 70.0, 75.0, 75.0],  # valid
    [200.0, 200.0, 201.0, 201.0]  # invalid
], dtype=torch.float32)

right_bbox_1 = torch.tensor([
    [12.0, 22.0, 17.0, 27.0],  # valid
    [32.0, 42.0, 33.0, 43.0],  # valid
    [52.0, 62.0, 52.5, 62.5],  # invalid
    [102.0, 102.0, 104.0, 104.0]  # valid
], dtype=torch.float32)

right_bbox_2 = torch.tensor([
    [6.0, 6.0, 11.0, 11.0],  # valid
    [21.0, 21.0, 21.5, 21.5],  # invalid
    [71.0, 71.0, 76.0, 76.0],  # valid
    [201.0, 201.0, 202.0, 202.0]  # invalid
], dtype=torch.float32)

left_results = [
    MockDetectionResult(left_bbox_1),
    MockDetectionResult(left_bbox_2)
]

right_results = [
    MockDetectionResult(right_bbox_1),
    MockDetectionResult(right_bbox_2)
]

ARGS = {
    'left_results': left_results,
    'right_results': right_results
}
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 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
remove_too_right_proposals Ran
ran; our contract draft was wrong, not the code
zju3dv/disprcnn/disprcnn/modeling/pointnet_module/point_rcnn/lib/net/point_rcnn.py
code served (permissive licence) · get_code("7fa6138ba5e0cabd")
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 torch
import numpy as np

torch.manual_seed(0)

# Create a mock result object that has a bbox attribute and supports indexing
class MockDetectionResult:
    def __init__(self, bbox):
        self.bbox = bbox
    
    def __getitem__(self, indices):
        # Support boolean indexing to filter by keep mask
        new_bbox = self.bbox[indices]
        return MockDetectionResult(new_bbox)

# Create sample bounding boxes for left and right results
# bbox shape: (num_detections, 4) where columns are [x1, y1, x2, y2] or similar
# We use shape (3, 4) for 3 detections with 4 coordinates each

left_bbox_1 = torch.tensor([
    [5.0, 1.0, 10.0, 15.0],
    [8.0, 2.0, 12.0, 16.0],
    [0.0, 3.0, 14.0, 17.0]
], dtype=torch.float32)

right_bbox_1 = torch.tensor([
    [3.0, 1.0, 9.0, 15.0],
    [2.0, 2.0, 11.0, 16.0],
    [1.0, 3.0, 13.0, 17.0]
], dtype=torch.float32)

left_bbox_2 = torch.tensor([
    [6.0, 4.0, 11.0, 18.0],
    [0.0, 5.0, 13.0, 19.0]
], dtype=torch.float32)

right_bbox_2 = torch.tensor([
    [4.0, 4.0, 10.0, 18.0],
    [2.0, 5.0, 12.0, 19.0]
], dtype=torch.float32)

# Create left and right result objects
left_result_1 = MockDetectionResult(left_bbox_1)
right_result_1 = MockDetectionResult(right_bbox_1)

left_result_2 = MockDetectionResult(left_bbox_2)
right_result_2 = MockDetectionResult(right_bbox_2)

# Package into lists of results (batch)
left_results = [left_result_1, left_result_2]
right_results = [right_result_1, right_result_2]

ARGS = {
    "left_results": left_results,
    "right_results": right_results
}
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, 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

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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 2026-11-21.

Repositories linked to this paper

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

Abstract

In this paper, we propose a novel system named Disp R-CNN for 3D object detection from stereo images. Many recent works solve this problem by first recovering a point cloud with disparity estimation and then apply a 3D detector. The disparity map is computed for the entire image, which is costly and fails to leverage category-specific prior. In contrast, we design an instance disparity estimation network (iDispNet) that predicts disparity only for pixels on objects of interest and learns a category-specific shape prior for more accurate disparity estimation. To address the challenge from scarcity of disparity annotation in training, we propose to use a statistical shape model to generate dense disparity pseudo-ground-truth without the need of LiDAR point clouds, which makes our system more widely applicable. Experiments on the KITTI dataset show that, even when LiDAR ground-truth is not available at training time, Disp R-CNN achieves competitive performance and outperforms previous state-of-the-art methods by 20% in terms of average precision.

For agents

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

get_harvested_code_for_paper("2004.03572")
get_code_for_paper("2004.03572")
have("2004.03572")

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

curl https://syntology.ai/api/ran/2004.03572.json

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

[![Syntology run record](https://syntology.ai/api/ran/2004.03572.svg)](https://syntology.ai/paper/2004.03572)

Connect an agent — have() is free.