SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1803.02276 · 2018

GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 2 have no contract verdict; 1 failure of our instrument; 4 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/1803.02276.json · badge: /api/ran/1803.02276.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
yzcjtr/GeoNet canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
FunctionStatusWhere it lives
compute_errors Ran
ran, violated the contract we drafted
yzcjtr/GeoNet/kitti_eval/depth_evaluation_utils.py
code served (permissive licence) · get_code("d83582de454a0b40")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch

torch.manual_seed(0)

# Create small numpy arrays for ground truth and predictions
# These should be 1D arrays of depth values (common for depth estimation tasks)
gt = np.array([1.0, 2.0, 3.0, 4.0, 5.0], dtype=np.float32)
pred = np.array([1.1, 1.9, 3.2, 3.8, 5.1], dtype=np.float32)

ARGS = {
    "gt": gt,
    "pred": pred,
}
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 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
euler2mat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yzcjtr/GeoNet/utils.py
code served (permissive licence) · get_code("46a3f7669dcd8ad5")
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 EagerTensor, ndim 4, dtype family float, shape (2, 1, 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
import tensorflow as tf

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

# Create small test tensors for euler angles
# Shape: [B, N] where B=2 (batch size), N=1 (deprecated dimension)
B = 2
N = 1

# Create euler angles in radians, shaped [B, N]
z = tf.constant([[0.5], [1.0]], dtype=tf.float32)  # shape [2, 1]
y = tf.constant([[0.3], [0.7]], dtype=tf.float32)  # shape [2, 1]
x = tf.constant([[0.2], [0.9]], dtype=tf.float32)  # shape [2, 1]

ARGS = {
    'z': z,
    'y': y,
    'x': x,
}
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 EagerTensor, ndim 4, dtype family float, shape (2, 1, 3, 3) — shape (2, 1, 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
load_gt_disp_kitti Ran
ran on a synthesized input
yzcjtr/GeoNet/kitti_eval/depth_evaluation_utils.py
code served (permissive licence) · get_code("6f8a54ff22b913e2")
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 list, 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 cv2
import tempfile
import os

# Create a temporary directory structure to simulate the KITTI dataset
temp_dir = tempfile.mkdtemp()
training_dir = os.path.join(temp_dir, 'training', 'disp_noc_0')
os.makedirs(training_dir, exist_ok=True)

# Create 200 dummy disparity images
for i in range(200):
    # Create a small dummy disparity image (uint16 format, as KITTI uses)
    # The function reads with cv2.imread(..., -1) which preserves the original format
    disp_data = np.random.randint(0, 5000, size=(32, 64), dtype=np.uint16)
    filename = os.path.join(training_dir, str(i).zfill(6) + '_10.png')
    cv2.imwrite(filename, disp_data)

ARGS = {
    'path': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 3, dtype family float — a list 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
pose_vec2mat Ran
ran on a synthesized input
yzcjtr/GeoNet/utils.py
code served (permissive licence) · get_code("82c27adf53eb9d69")
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 EagerTensor, ndim 3, dtype family float, shape (2, 4, 4) 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
import tensorflow as tf

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

# Create small test inputs for pose_vec2mat
# vec should be [B, 6] where B is batch size
# The 6 values are: tx, ty, tz, rx, ry, rz (translation and rotation angles)

batch_size = 2
vec = tf.constant([
    [0.1, 0.2, 0.3, 0.01, 0.02, 0.03],  # First sample: small translations and rotations
    [0.05, 0.15, 0.25, 0.005, 0.015, 0.025]  # Second sample
], dtype=tf.float32)

ARGS = {
    "vec": vec
}
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 EagerTensor, ndim 3, dtype family float, shape (2, 4, 4) — shape (2, 4, 4) 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
convert_disps_to_depths_kitti Not yet run
unverified (harvested, no recorded run)
yzcjtr/GeoNet/kitti_eval/depth_evaluation_utils.py
code served (permissive licence) · get_code("cf98eb3840092327")
How it ran

no invocation recorded.

load_image_sequence Not yet run
unverified (harvested, no recorded run)
yzcjtr/GeoNet/geonet_test_pose.py
code served (permissive licence) · get_code("8fc3bf3cf5e6a6ba")
How it ran

no invocation recorded.

pixel2cam Not yet run
unverified (harvested, no recorded run)
yzcjtr/GeoNet/utils.py
code served (permissive licence) · get_code("883ac80de85205ef")
How it ran

no invocation recorded.

unpack_image_sequence Not yet run
unverified (harvested, no recorded run)
yzcjtr/GeoNet/geonet_test_flow.py
code served (permissive licence) · get_code("65362a4a8daa0fdd")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 5 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 5 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.

FunctionAlso present inEarliest-datedOthers named
compute_errors 24 other papers
1 dated earlier, 23 later
1609.03677 (2016-09)
Unsupervised Monocular Depth Estimation with Left-Right Consistency
1806.01260 (2018-06)
Digging Into Self-Supervised Monocular Depth Estimation
1909.09051 (2019-09)
Self-Supervised Monocular Depth Hints
and 21 more not named here
convert_disps_to_depths_kitti 2 other papers
1 dated earlier, 1 later
1609.03677 (2016-09)
Unsupervised Monocular Depth Estimation with Left-Right Consistency
1909.09115 (2019-09)
Self-Supervised Learning of Depth and Motion Under Photometric Inconsistency
load_gt_disp_kitti 2 other papers
1 dated earlier, 1 later
1609.03677 (2016-09)
Unsupervised Monocular Depth Estimation with Left-Right Consistency
1909.09115 (2019-09)
Self-Supervised Learning of Depth and Motion Under Photometric Inconsistency
euler2mat 1 other paper
1 later
1909.09115 (2019-09)
Self-Supervised Learning of Depth and Motion Under Photometric Inconsistency
pixel2cam 1 other paper
1 later
1805.09817 (2018-05)
Stereo Magnification: Learning View Synthesis using Multiplane Images

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

We propose GeoNet, a jointly unsupervised learning framework for monocular depth, optical flow and ego-motion estimation from videos. The three components are coupled by the nature of 3D scene geometry, jointly learned by our framework in an end-to-end manner. Specifically, geometric relationships are extracted over the predictions of individual modules and then combined as an image reconstruction loss, reasoning about static and dynamic scene parts separately. Furthermore, we propose an adaptive geometric consistency loss to increase robustness towards outliers and non-Lambertian regions, which resolves occlusions and texture ambiguities effectively. Experimentation on the KITTI driving dataset reveals that our scheme achieves state-of-the-art results in all of the three tasks, performing better than previously unsupervised methods and comparably with supervised ones.

For agents

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

get_harvested_code_for_paper("1803.02276")
get_code_for_paper("1803.02276")
have("1803.02276")

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

curl https://syntology.ai/api/ran/1803.02276.json

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

[![Syntology run record](https://syntology.ai/api/ran/1803.02276.svg)](https://syntology.ai/paper/1803.02276)

Connect an agent — have() is free.