Synthetic-Child: An AIGC-Based Synthetic Data Pipeline for Privacy-Preserving Child Posture Estimation

Taowen Zeng

cs.CVMar 3, 2026

Synthetic-Child: An AIGC-Based Synthetic Data Pipeline for Privacy-Preserving Child Posture Estimation

Authors: Taowen Zeng

Organizations: Independent Researcher

Abstract

Accurate child posture estimation is critical for AI-powered study companion devices, yet collecting large-scale annotated datasets of children is both expensive and ethically prohibitive due to privacy concerns. We present Synthetic-Child, an AIGC-based synthetic data pipeline that produces photorealistic child posture training images with ground-truth-projected keypoint annotations, requiring zero real child photographs. The pipeline comprises four stages: (1) a programmable 3D child body model (SMPL-X) in Blender generates diverse desk-study poses with IK-constrained anatomical plausibility and automatic COCO-format ground-truth export; (2) a custom PoseInjectorNode feeds 3D-derived skeletons into a dual ControlNet (pose + depth) conditioned on FLUX-1 Dev, synthesizing 12,000 photorealistic images across 10 posture categories with low annotation drift; (3) ViTPose-based confidence filtering and targeted augmentation remove generation failures and improve robustness; (4) RTMPose-M (13.6M params) is fine-tuned on the synthetic data and paired with geometric feature engineering and a lightweight MLP for posture classification, then quantized to INT8 for real-time edge deployment. On a real-child test set (n~300), the FP16 model achieves 71.2 AP -- a +12.5 AP improvement over the COCO-pretrained adult-data baseline at identical model capacity. After INT8 quantization the model retains 70.4 AP while running at 22 FPS on a 0.8-TOPS Rockchip RK3568 NPU. In a single-subject controlled comparison with a commercial posture corrector, our system achieves substantially higher recognition rates across most tested categories and responds ~1.8x faster on average. These results demonstrate that carefully designed AIGC pipelines can substantially reduce dependence on real child imagery while achieving deployment-ready accuracy, with potential applications to other privacy-sensitive domains.

Figures & tables

Figure 1: Overview of the Synthetic-Child pipeline. Stage 1 : A child-specific SMPL-X model in Blender generates diverse desk-study poses with IK constraints, exporting rendered images, depth maps, and COCO-format keypoint annotations. Stage 2 : A custom PoseInjectorNode feeds ground-truth skeletons into a dual ControlNet (pose + depth) conditioned on FLUX-1 Dev, producing 12,000 photorealistic child images with low annotation drift. Stage 3 : Three automated quality gates—confidence filtering (ViTPose-H), spatial fidelity checking, and category consistency verification—remove generation failures and yield 11,900 curated images from an initial pool of 12,000; online augmentation further increases robustness. Stage 4 : RTMPose-M is fine-tuned on the synthetic data; geometric feature engineering and a lightweight MLP classify postures; the model is quantized to INT8 for real-time inference on an edge NPU.

ID	Category	Description
0	correct_posture	Upright spine, shoulders level, eyes forward
1	bowed_head	Moderate forward head tilt ( ${\sim}15$ – $30^{\circ}$ )
2	very_bowed	Severe forward flexion ( $>30^{\circ}$ )
3	lean_desk	Upper body collapsed onto desk surface
4	lean_left	Lateral trunk lean to the left
5	lean_right	Lateral trunk lean to the right

Table 1: The

C{=}10

posture categories generated in Stage 1. Each template specifies a canonical joint configuration; stochastic perturbation (Eq. 2 ) generates diverse variants within each category.

Figure 2: Illustration of the Stage 1 $\to$ Stage 2 pipeline for two representative samples viewed from a frontal desk-mounted camera perspective. Row 1 : correct upright posture; Row 2 : upper body collapsed onto desk ( lean_desk ). Each row shows: (a) Blender render with overlaid 3D skeleton, (b) depth map rendered from the Blender Z-buffer, (c) COCO-17 OpenPose skeleton deterministically rendered by PoseInjectorNode from the ground-truth JSON, and (d) photorealistic image generated by FLUX-1 conditioned on both control signals (pose + depth). The stark visual contrast between (a) and (d) demonstrates the pipeline’s ability to bridge the sim-to-real gap while faithfully preserving the input pose.

Training Data	Model	Params	AP	[email protected]	FPS
		(M)			(NPU)
COCO pre-trained	RTMPose-M	13.6	58.7	71.4	18
Synthetic-Child	RTMPose-M	13.6	71.2	84.5	18
Synthetic-Child (INT8 $\dagger$ )	RTMPose-M	13.6	70.4	83.7	22

Table 2: Keypoint estimation accuracy on the real-child test set (

n{\approx}300

). All baseline and proposed models use the RTMPose-M architecture (13.6M params). “

\dagger

” denotes INT8 quantized model; all other rows use FP16 precision. All models are deployed on the same RK3568 NPU. Best results are in bold .

Posture Behavior	Angle/Condition	$n$	Ours	Commercial	$\boldsymbol{\Delta}$
Head turning	$15^{\circ}$ – $50^{\circ}$ ( $5^{\circ}$ steps)	24	87.5%	83.3%	+4.2%
Head too low	$20^{\circ}$ – $60^{\circ}$ ( $10^{\circ}$ steps)	15	86.7%	6.7%	+80.0%
Lateral leaning	$10^{\circ}$ – $40^{\circ}$ ( $5^{\circ}$ steps)	21	66.7%	47.6%	+19.1%
Body rotation	$15^{\circ}$ – $50^{\circ}$ ( $5^{\circ}$ steps)	24	91.7%	—	—
Lying on desk	Fully collapsed	3	100%	100%	0%

Table 3: Pilot recognition rate (RR) comparison with a commercial RGB-based posture corrector (single subject, single environment). RR is the proportion of test instances where the device triggers a correct alert within 30 s. “—” indicates the device does not support this detection category.

Posture Behavior	$n$	Ours (s)	Commercial (s)	Speedup
Head turning	24	4.71	10.20	$2.16\times$
Head too low	15	9.08	10.00	$1.10\times$
Lateral leaning	21	6.50	13.70	$2.11\times$
Body rotation	24	5.80	—	—
Lying on desk	3	6.33	10.00	$1.58\times$
Weighted avg. (shared)	63	6.42	11.31	$1.76\times$

Table 4: Average response time

T_{\text{avg}}

(seconds) from posture onset to alert trigger. Per-category averages are computed over successfully detected instances;

n

denotes the total number of test instances per category (same as Table 3 ). Lower is better. “—” indicates the device does not support this category. The weighted average is computed over shared categories only.

Configuration	AP	[email protected]	$\boldsymbol{\Delta}$ AP
Full Synthetic-Child pipeline	71.2	84.5	—
(a) Synthesis configuration
w/o depth ControlNet ( $w_{2}{=}0$ )	68.1	81.5	$-$ 3.1
(b) Quality control
w/o ViTPose confidence filtering	67.8	81.2	$-$ 3.4
w/o data augmentation	69.1	82.6	$-$ 2.1

Table 5: Ablation study on key pipeline components. All variants use RTMPose-M evaluated on the real-child test set.

\Delta

AP denotes the change relative to the full pipeline.

Panel of Figure 3 — Figure 3: Qualitative comparison on challenging child postures from the real-child test set, selected for their difficulty under the COCO-pretrained baseline. Each pair shows the baseline output ( left : estimated keypoint skeleton only) and the full Synthetic-Child pipeline output ( right : keypoints + detection bounding box + MLP classification label with confidence score). (a, b) Two subjects leaning onto the desk surface with ${\geq}$ 0.97 confidence, a common desk-study posture where children’s compressed upper body and foreshortened limbs diverge markedly from adult training distributions. (c) Rightward lateral trunk lean (0.96 confidence). (d) Correct upright posture (0.91 confidence), included to show that the pipeline correctly identifies proper sitting as non-deviant rather than triggering false alerts. The baseline provides keypoint estimation only and cannot classify posture; our pipeline enables category-level posture alerts with high confidence.

Human Pose Estimation

Human pose estimation recovers people’s body keypoints, 3D skeletons, or meshes from images, video, and other sensor data.

See all 74 papers

Patient Pose Assessment Using a CT-Based Framework for Synthetic Data Generation
Aug 6, 2026Manuel Laufer, Dominik Mairhöfer, Malte Sieren +7Synthetic Data Pretraining Medical Imaging
Institute for Neuro- and Bioinformatics, University of Lübeck, Germany · University Medical Center Schleswig-Holstein, Lübeck, Germany · IMAGE Information Systems Europe GmbH, Rostock, Germany +1
PoseDreamer: Scalable and Photorealistic Human Data Generation Pipeline with Diffusion Models
Mar 30, 2026Lorenza Prospero, Orest Kupyn, Ostap Viniavskyi +2Synthetic Data Generation Human Mesh Recovery
The Podium Institute for Sports Medicine and Technology University of Oxford Oxford, UK · Visual Geometry Group University of Oxford Oxford, UK · Ukrainian Catholic University Lviv, Ukraine
Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AI
Jul 14, 2026Martina Radoynova, Samuel Pantze, Trina De +2Synthetic-to-Real Domain Adaptation Synthetic Data Generation
Center for Advanced Systems Understanding (CASUS), Görlitz, Germany · Helmholtz-Zentrum Dresden-Rossendorf e. V. (HZDR), Dresden, Germany · Institute of Computer Science, University of Wrocław, Wrocław, Poland +1

Aug 6, 2026J/K move · Enter open · S save

Patient Pose Assessment Using a CT-Based Framework for Synthetic Data Generation

Manuel Laufer, Dominik Mairhöfer, Malte Sieren, Hauke Gerdes+6

Institute for Neuro- and Bioinformatics, University of Lübeck, Germany · University Medical Center Schleswig-Holstein, Lübeck, Germany · IMAGE Information Systems Europe GmbH, Rostock, Germany+1

Synthetic Data Pretraining Medical Imaging Synthetic Data Generation Medical Image Synthesis

Open paper AskPDF arXiv

An adequate diagnostic quality of radiographs is essential for reliable diagnoses and treatment planning. The patient's pose during radiography is one of the most important factors determining the diagnostic quality. Since patient positioning is difficult and not standardized, an automated AI-based approach using depth images to automatically assess the patient's pose before the radiograph has been taken would be helpful. Due to regulatory hurdles, however, it is difficult in practice to acquire the required depth images and corresponding radiographs. In this paper, we present a framework that can generate such training data synthetically from Computed Tomography scans. We further show that by pretraining on our generated synthetic dataset consisting of 3077 image pairs of upper ankle joints, the pose assessment of real upper ankle joints can be improved by up to 11 percentage points.

Abstract

Human Pose Estimation

Explore similar work

Patient Pose Assessment Using a CT-Based Framework for Synthetic Data Generation

PoseDreamer: Scalable and Photorealistic Human Data Generation Pipeline with Diffusion Models

Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AI