SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2107.11291 · ICCV · 2021

Human Pose Regression with Residual Log-likelihood Estimation

Bo Pang, Can Wang, Cewu Lu, Jiefeng Li, Ailing Zeng, Wentao Liu, Siyuan Bian

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 0 failures of our instrument; 1 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/2107.11291.json · badge: /api/ran/2107.11291.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
sithu31296/pose-estimation community (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
ligaoqi2/res-loglikelihood-regression-with-COCO-video-demo community
Papers with Code archive
2 ran · 0 no recorded run
FunctionStatusWhere it lives
RLELoss Ran
ran on a synthesized input · constructed an object; no result was computed
ligaoqi2/res-loglikelihood-regression-with-COCO-video-demo/rlepose/models/criterion.py
pointer only (licence: NONE) · get_code("05bc1069a40fc721")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ligaoqi2/res-loglikelihood-regression-with-COCO-video-demo 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.

Registry Ran
ran on a synthesized input · constructed an object; no result was computed
ligaoqi2/res-loglikelihood-regression-with-COCO-video-demo/rlepose/models/criterion.py
pointer only (licence: NONE) · get_code("99274af8ba3f2642")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ligaoqi2/res-loglikelihood-regression-with-COCO-video-demo 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.

draw_coco_keypoints Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/utils.py
code served (permissive licence) · get_code("2a3363878b0ddfdf")
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 3, dtype family int, shape (100, 100, 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 cv2
import numpy as np

# Create a small test image (100x100 RGB)
img = np.zeros((100, 100, 3), dtype=np.uint8)

# Create sample keypoints
# Each keypoint is [x, y, v] where v is visibility (2 = visible)
# We have 2 instances with 5 keypoints each (simulating a small pose model)
keypoints = [
    [
        [20, 30, 2],  # keypoint 1
        [40, 35, 2],  # keypoint 2
        [60, 40, 2],  # keypoint 3
        [25, 60, 2],  # keypoint 4
        [55, 65, 2],  # keypoint 5
    ],
    [
        [70, 20, 2],  # keypoint 1
        [80, 25, 2],  # keypoint 2
        [90, 30, 2],  # keypoint 3
        [75, 50, 2],  # keypoint 4
        [85, 55, 2],  # keypoint 5
    ],
]

# Define skeleton connections (pairs of keypoint indices, 1-indexed)
# Connects: 1-2, 2-3, 1-4, 4-5, 3-5
skeletons = [
    (1, 2),
    (2, 3),
    (1, 4),
    (4, 5),
    (3, 5),
]

ARGS = {
    "img": img,
    "keypoints": keypoints,
    "skeletons": skeletons,
}
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 ndarray, ndim 3, dtype family int, shape (100, 100, 3) — shape (100, 100, 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
get_dir Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/utils.py
code served (permissive licence) · get_code("54ada554be189976")
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

# Analyze the function:
# - src_point: used with indexing [0] and [1], so it's a sequence/array with at least 2 elements
# - rot: used in arithmetic (np.pi * rot / 180), so it's a numeric scalar

src_point = np.array([3.0, 4.0])
rot = 45.0

ARGS = {
    "src_point": src_point,
    "rot": rot
}
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, 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
get_final_preds Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/decode.py
code served (permissive licence) · get_code("2737d84e3576e90f")
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 3, dtype family int, shape (2, 4, 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
import numpy as np
import math

torch.manual_seed(0)

# Create small heatmaps tensor
# Shape: (B, C, H, W) = (2, 4, 8, 8)
# Values between 0 and 1 to simulate heatmap probabilities
heatmaps = torch.rand(2, 4, 8, 8, dtype=torch.float32)

# Create boxes tensor
# Shape: (B, 4) where each row is [center_x, center_y, scale_x, scale_y]
boxes = torch.tensor([
    [64.0, 64.0, 0.5, 0.5],
    [128.0, 128.0, 0.6, 0.6]
], dtype=torch.float32)

ARGS = {
    "heatmaps": heatmaps,
    "boxes": boxes
}
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 ndarray, ndim 3, dtype family int, shape (2, 4, 2) — shape (2, 4, 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_max_preds Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/decode.py
code served (permissive licence) · get_code("ab19d6d591eec189")
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 3, dtype family float, shape (2, 3, 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 numpy as np

# Construct valid arguments for get_max_preds
# The function expects:
# - heatmaps: np.ndarray with shape (B, C, H, W)
#   where B is batch size, C is number of channels,
#   H and W are height and width

# Create a small batch of heatmaps
B = 2  # batch size
C = 3  # number of channels
H = 4  # height
W = 5  # width

# Generate random heatmap values between 0 and 1
heatmaps = np.random.rand(B, C, H, W).astype(np.float32)

# Ensure some non-zero values for meaningful results
heatmaps[0, 0, 1, 2] = 0.9
heatmaps[1, 1, 3, 4] = 0.85

ARGS = {
    "heatmaps": heatmaps
}
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 3, dtype family float, shape (2, 3, 2) — shape (2, 3, 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_simdr_final_preds Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/decode.py
code served (permissive licence) · get_code("7a838a31b94b6740")
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 3, dtype family int, shape (2, 3, 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
import numpy as np

torch.manual_seed(0)

# Create pred_x: shape (batch_size=2, num_joints=3, output_size=4)
# The function calls .softmax(dim=2) and .max(dim=2)[-1]
pred_x = torch.randn(2, 3, 4)

# Create pred_y: shape (batch_size=2, num_joints=3, output_size=5)
# Using different size than pred_x for clarity
pred_y = torch.randn(2, 3, 5)

# Create boxes: shape (batch_size=2, 4)
# First 2 columns are center (x, y), last 2 columns are scale (sx, sy)
boxes = torch.tensor([
    [100.0, 150.0, 0.5, 0.6],
    [200.0, 250.0, 0.7, 0.8]
], dtype=torch.float32)

# image_size is a tuple used in transform_preds
image_size = (64, 48)

ARGS = {
    "pred_x": pred_x,
    "pred_y": pred_y,
    "boxes": boxes,
    "image_size": image_size
}
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 ndarray, ndim 3, dtype family int, shape (2, 3, 2) — shape (2, 3, 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
letterbox Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/boxes.py
code served (permissive licence) · get_code("128c6d6cc48dcf20")
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 3, dtype family int, shape (480, 640, 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 cv2
import numpy as np

# Create a small test image (BGR format as expected by cv2)
# Using a 480x640 image to have distinct dimensions
img = np.random.randint(0, 255, (480, 640, 3), dtype=np.uint8)

# Define ARGS dict with all parameters
ARGS = {
    "img": img,
    "new_shape": (640, 640)
}
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 ndarray, ndim 3, dtype family int, shape (480, 640, 3) — shape (480, 640, 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
scale_boxes Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/boxes.py
code served (permissive licence) · get_code("55e839c9dc92868b")
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, 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 torch

torch.manual_seed(0)

# Create a small batch of bounding boxes
# boxes shape: (batch_size, num_boxes_info)
# The function uses boxes[:, ::2], boxes[:, 1::2], boxes[:, :4]
# So we need at least 4 columns for boxes
boxes = torch.tensor([
    [10.0, 20.0, 50.0, 60.0],
    [15.0, 25.0, 55.0, 65.0]
], dtype=torch.float32)

# Original image shape (Height, Width)
orig_shape = (100, 120)

# New shape after scaling (Height, Width)
new_shape = (200, 240)

ARGS = {
    "boxes": boxes,
    "orig_shape": orig_shape,
    "new_shape": new_shape
}
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, 4) — shape (2, 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
xywh2xyxy Ran
ran on a synthesized input
sithu31296/pose-estimation/pose/utils/boxes.py
code served (permissive licence) · get_code("e31e0122736a3263")
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, 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 torch

torch.manual_seed(0)

# The function xywh2xyxy takes a tensor x with shape (batch_size, 4)
# where each row is [x_center, y_center, width, height]
# and converts it to [x_min, y_min, x_max, y_max] format

x = torch.tensor([
    [10.0, 20.0, 4.0, 6.0],
    [30.0, 40.0, 8.0, 10.0]
], dtype=torch.float32)

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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
draw_keypoints Not yet run
unverified (harvested, no recorded run)
sithu31296/pose-estimation/pose/utils/utils.py
code served (permissive licence) · get_code("8ea8636e78b2920a")
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, 3 are also present in other papers whose code we hold (3 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
Registry
brick of this paper's assembly: helper
5 other papers
2 dated earlier, 3 later · 4 only as an assembly brick
2104.00885 (2021-04)
Adaptive Class Suppression Loss for Long-Tail Object Detection
2104.08683 (2021-04)
Self-Supervised Pillar Motion Learning for Autonomous Driving
2108.10312 (2021-08)
Exploring Simple 3D Multi-Object Tracking for Autonomous Driving
and 2 more not named here
letterbox 1 other paper
1 dated earlier
2104.14294 (2021-04)
Emerging Properties in Self-Supervised Vision Transformers
scale_boxes 1 other paper
1 dated earlier
2104.14294 (2021-04)
Emerging Properties in Self-Supervised Vision Transformers

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

Heatmap-based methods dominate in the field of human pose estimation by modelling the output distribution through likelihood heatmaps. In contrast, regressionbased methods are more efficient but suffer from inferior performance. In this work, we explore maximum likelihood estimation (MLE) to develop an efficient and effective regression-based method. From the perspective of MLE, adopting different regression losses is making different assumptions about the output density function. A density function closer to the true distribution leads to a better regression performance. In light of this, we propose a novel regression paradigm with Residual Log-likelihood Estimation (RLE) to capture the underlying output distribution. Concretely, RLE learns the change of the distribution instead of the unreferenced underlying distribution to facilitate the training process. With the proposed reparameterization design, our method is compatible with off-theshelf flow models. The proposed method is effective, efficient and flexible. We show its potential in various human pose estimation tasks with comprehensive experiments. Compared to the conventional regression paradigm, regression with RLE bring 12.4 mAP improvement on MSCOCO without any test-time overhead. Moreover, for the first time, especially on multi-person pose estimation, our regression method is superior to the heatmap-based methods. Our code is available at https://github.com/Jeff-sjtu/resloglikelihood-regression.

For agents

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

get_harvested_code_for_paper("2107.11291")
get_code_for_paper("2107.11291")
have("2107.11291")

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

curl https://syntology.ai/api/ran/2107.11291.json

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

[![Syntology run record](https://syntology.ai/api/ran/2107.11291.svg)](https://syntology.ai/paper/2107.11291)

Connect an agent — have() is free.