SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.01193 · 2019

Triangulation Learning Network: from Monocular to Stereo 3D Object Detection

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 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 7 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/1906.01193.json · badge: /api/ran/1906.01193.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
Zengyi-Qin/TLNet community (classifier label)
classifier basis not recorded on the link
7 ran · 2 no recorded run
FunctionStatusWhere it lives
box_3d_to_anchor Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/box_3d_encoder.py
code served (permissive licence) · get_code("3198a0be3b6890f8")
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 (4, 6) 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

# Create sample 3D boxes in the format [x, y, z, l, w, h, ry]
# where x, y, z are center coordinates
#       l, w, h are length, width, height
#       ry is rotation around y-axis in radians

boxes_3d_sample = np.array([
    [0.5, 1.0, 2.0, 3.5, 2.5, 1.8, 0.0],        # box with no rotation
    [1.5, 2.0, 3.5, 4.0, 3.0, 2.0, np.pi/2],    # box with 90 degree rotation
    [2.5, 1.5, 1.0, 2.8, 2.2, 1.5, np.pi/4],    # box with 45 degree rotation
    [3.0, 0.5, 2.5, 3.2, 2.8, 1.9, 0.0],        # another box with no rotation
])

ARGS = {
    "boxes_3d": boxes_3d_sample,
    "ortho_rotate": False,
}
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 (4, 6) — shape (4, 6) 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_stereo_calibration Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/calib_utils.py
code served (permissive licence) · get_code("2b65c8ed7c2c4993")
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 StereoCalibrationData. 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

# Create valid 3x4 camera matrices for left and right cameras
# These should be realistic projection matrices with proper structure

# Left camera matrix (3x4)
left_cam_mat = np.array([
    [500.0,   0.0, 320.0,   0.0],
    [  0.0, 500.0, 240.0,   0.0],
    [  0.0,   0.0,   1.0,   0.0]
], dtype=np.float64)

# Right camera matrix (3x4) - similar to left but with baseline offset
# The translation component (last column) differs to encode the baseline
right_cam_mat = np.array([
    [500.0,   0.0, 320.0, -2500.0],
    [  0.0, 500.0, 240.0,    0.0],
    [  0.0,   0.0,   1.0,    0.0]
], dtype=np.float64)

ARGS = {
    "left_cam_mat": left_cam_mat,
    "right_cam_mat": right_cam_mat,
}
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 StereoCalibrationData — a StereoCalibrationData output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a StereoCalibrationData 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
krt_from_p Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/calib_utils.py
code served (permissive licence) · get_code("49bd82d9ef4615f3")
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

# Create a valid 3x4 projection matrix P
# P = K[R|t] where K is 3x3, R is 3x3, t is 3x1
# We'll construct a simple valid projection matrix

# Create a simple intrinsic matrix K
K = np.array([
    [500.0, 0.0, 320.0],
    [0.0, 500.0, 240.0],
    [0.0, 0.0, 1.0]
])

# Create a simple rotation matrix R (identity for simplicity)
R = np.eye(3)

# Create a simple translation vector t
t = np.array([[1.0], [2.0], [3.0]])

# Construct P = K[R|t]
P = np.hstack([K @ R, K @ t])

# P should be 3x4
assert P.shape == (3, 4), f"P shape is {P.shape}, expected (3, 4)"

ARGS = {
    "p": P,
    "fsign": 1
}
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
object_label_to_box_3d Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/box_3d_encoder.py
code served (permissive licence) · get_code("b58ba17ac58a188e")
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 1, dtype family float, shape (7,) 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

class ObjectLabel:
    """Stand-in class for ObjectLabel with required attributes"""
    def __init__(self):
        self.t = np.array([1.5, 2.3, 3.7])  # position [x, y, z]
        self.l = 4.2  # length
        self.w = 1.8  # width
        self.h = 1.5  # height
        self.ry = 0.5  # rotation around y-axis

obj_label = ObjectLabel()

ARGS = {
    "obj_label": obj_label
}
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 1, dtype family float, shape (7,) — shape (7,) 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
project_to_bev Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/anchor_projector.py
code served (permissive licence) · get_code("33c26f8c51c4dda9")
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 3, 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 tensorflow as tf

# Create sample anchors: N x [x, y, z, dim_x, dim_y, dim_z]
# Using 2 anchors with 6 coordinates each
anchors_np = np.array([
    [0.5, 1.0, 2.0, 0.4, 0.5, 0.6],
    [1.5, 1.5, 3.0, 0.3, 0.4, 0.5]
], dtype=np.float32)

# BEV extents: [[min_x, max_x], [min_z, max_z]]
bev_extents = [
    [0.0, 4.0],  # x extents
    [0.0, 5.0]   # z extents
]

ARGS = {
    "anchors": anchors_np,
    "bev_extents": bev_extents
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, 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
read_calibration Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/calib_utils.py
code served (permissive licence) · get_code("658a79905061825c")
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 FrameCalibrationData. 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 tempfile
import os
import csv
import numpy as np

# Create a temporary directory and a calibration file
temp_dir = tempfile.mkdtemp()

# Create a calibration file with the expected format
calib_file_path = os.path.join(temp_dir, '000000.txt')

# KITTI calibration format:
# P0: 3x4 matrix
# P1: 3x4 matrix
# P2: 3x4 matrix
# P3: 3x4 matrix
# R0_rect: 3x3 matrix
# Tr_velo_to_cam: 3x4 matrix

calibration_data = [
    "P0: 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0",
    "P1: 1.1 2.1 3.1 4.1 5.1 6.1 7.1 8.1 9.1 10.1 11.1 12.1",
    "P2: 1.2 2.2 3.2 4.2 5.2 6.2 7.2 8.2 9.2 10.2 11.2 12.2",
    "P3: 1.3 2.3 3.3 4.3 5.3 6.3 7.3 8.3 9.3 10.3 11.3 12.3",
    "R0_rect: 1.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0",
    "Tr_velo_to_cam: 1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0"
]

with open(calib_file_path, 'w') as f:
    for line in calibration_data:
        f.write(line + '\n')

ARGS = {
    'calib_dir': temp_dir,
    'img_idx': 0
}
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 FrameCalibrationData — a FrameCalibrationData output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a FrameCalibrationData 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
tf_project_to_image_space Ran
ran on a synthesized input
Zengyi-Qin/TLNet/avod/core/anchor_projector.py
code served (permissive licence) · get_code("a0851690ea0a7340")
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 3, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tensorflow as tf
import numpy as np

# Set random seed for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create sample anchors: shape [N, 6] where N=2
# Format: [x, y, z, dim_x, dim_y, dim_z]
anchors = tf.constant([
    [10.0, 5.0, 20.0, 2.0, 1.5, 3.0],
    [15.0, 8.0, 25.0, 2.5, 1.8, 3.5]
], dtype=tf.float32)

# Create stereo camera calibration P2 matrix: shape [3, 4]
# This is a typical camera intrinsic matrix combined with extrinsic parameters
stereo_calib_p2 = tf.constant([
    [721.5377, 0.0, 609.5593, 44.85728],
    [0.0, 721.5377, 172.854, 0.2163791],
    [0.0, 0.0, 1.0, 0.002745884]
], dtype=tf.float32)

# Create image shape: [height, width]
image_shape = tf.constant([376.0, 1242.0], dtype=tf.float32)

ARGS = {
    'anchors': anchors,
    'stereo_calib_p2': stereo_calib_p2,
    'image_shape': image_shape
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, 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
area Not yet run
unverified (harvested, no recorded run)
Zengyi-Qin/TLNet/avod/core/box_list_ops.py
code served (permissive licence) · get_code("d4fcd5271128b6e4")
How it ran

no invocation recorded.

height_width Not yet run
unverified (harvested, no recorded run)
Zengyi-Qin/TLNet/avod/core/box_list_ops.py
code served (permissive licence) · get_code("7f812aae8034ae0c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 8 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
area 1 other paper
1 later
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking

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

In this paper, we study the problem of 3D object detection from stereo images, in which the key challenge is how to effectively utilize stereo information. Different from previous methods using pixel-level depth maps, we propose employing 3D anchors to explicitly construct object-level correspondences between the regions of interest in stereo images, from which the deep neural network learns to detect and triangulate the targeted object in 3D space. We also introduce a cost-efficient channel reweighting strategy that enhances representational features and weakens noisy signals to facilitate the learning process. All of these are flexibly integrated into a solid baseline detector that uses monocular images. We demonstrate that both the monocular baseline and the stereo triangulation learning network outperform the prior state-of-the-arts in 3D object detection and localization on the challenging KITTI dataset.

For agents

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

get_harvested_code_for_paper("1906.01193")
get_code_for_paper("1906.01193")
have("1906.01193")

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

curl https://syntology.ai/api/ran/1906.01193.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.01193.svg)](https://syntology.ai/paper/1906.01193)

Connect an agent — have() is free.