Monocular Mesh Recovery and Body Measurement of Female Saanen Goats
Authors: Bo Jin, Shichao Zhao, Jin Lyu, Bin Zhang, Tao Yu, Liang An, Yebin Liu, Meili Wang
Organizations: College of Information Engineering, Northwest A&F University · Department of Electronic and Electrical Engineering, Southern University of Science and Technology · Beijing National Research Center for Information Science and Technology, Tsinghua University · Department of Automation, Tsinghua University · Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture · Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service · Shaanxi Engineering Research Center of Agricultural Information Intelligent Perception and Analysis
The lactation performance of Saanen dairy goats, renowned for their high milk yield, is intrinsically linked to their body size, making accurate 3D body measurement essential for assessing milk production potential, yet existing reconstruction methods lack goat-specific authentic 3D data. To address this limitation, we establish the FemaleSaanenGoat dataset containing synchronized eight-view RGBD videos of 55 female Saanen goats (6-18 months). Using multi-view DynamicFusion, we fuse noisy, non-rigid point cloud sequences into high-fidelity 3D scans, overcoming challenges from irregular surfaces and rapid movement. Based on these scans, we develop SaanenGoat, a parametric 3D shape model specifically designed for female Saanen goats. This model features a refined template with 41 skeletal joints and enhanced udder representation, registered with our scan data. A comprehensive shape space constructed from 48 goats enables precise representation of diverse individual variations. With the help of SaanenGoat model, we get high-precision 3D reconstruction from single-view RGBD input, and achieve automated measurement of six critical body dimensions: body length, height, chest width, chest girth, hip width, and hip height. Experimental results demonstrate the superior accuracy of our method in both 3D reconstruction and body measurement, presenting a novel paradigm for large-scale 3D vision applications in precision livestock farming.
Figures & tables
Figure 1: Pose variation examples in the Saanen dataset. (a) standing, (b) walking, (c) head turning, (d) looking backward, and (e) head lowering. Top row: fusion meshes after denoising; Middle row: manually labeled 3D surface keypoints; Bottom row: source depth images of one view.
Figure 2: Data Processing Pipeline. We capture RGBD data of Saanen dairy goats using top-view and seven surrounding cameras. Multi-view dynamic fusion generates geometrically accurate 3D goat models. The Saanen Goat Initialization Template is registered to the scanned data, followed by pose normalization to T-pose for shape space construction.
Figure 3: Examples of scan results for Saanen dairy goats of different ages and body types included in the dataset.
Figure 4: Comparison between the average model of our SaanenGoat (top) and the SMAL (bottom). Learned from real goat scan meshes, the SaanenGoat model provides a more accurate fit to the real goat’s anatomy than SMAL.
Figure 5: Visualization of the shape space for Saanen goats. The central grid represents the average shape, while the surrounding variations demonstrate different shape changes in the model. We visualize the first four principal components with deviations of ±2 standard deviations.
Figure 6: Comparison of model fitting performance on scanned goat data: ground truth mesh, SMAL, SMAL+, and SaanenGoat fittings with corresponding distance error maps. Distance visualizations use blue-to-red gradients indicating minimal-to-maximal fitting errors.
Chamfer Distance
Mesh-to-Scan
mean
std
Mean
std
SMAL
36.72
4.76
31.53
3.31
SMAL+
31.18
3.26
25.14
2.79
Ours
10.65
0.62
7.02
0.46
Table 1: Fitting Results on the In-Shape testset. Compared to both the neutral template SMAL model and SMAL+ model. Errors in mm.
Chamfer Distance
Mesh-to-Scan
mean
std
Mean
std
SMAL
39.68
5.14
33.71
4.35
SMAL+
32.46
3.75
26.32
2.87
Ours
15.30
1.83
11.29
1.76
Table 2: Out-Shape testset results. Comparison of reconstruction errors between our proposed model and the SMAL and SMAL+ baseline models. Errors in mm.
Figure 7: The methodology for body size measurement.BL: Body Length, BH: Body Height, HH: Hip Height, CW: Chest Width, HW: Hip Width, CG: Chest Girth. Left: manual measurement. Middle: key point-based measurements derived from the SaanenGoat model. Right: key point-based measurements derived from the SMAL model.
Figure 8: Body measurement comparison between SMAL, SMAL+, and our SaanenGoat model. Our model achieves closer agreement with manual measurements by more accurately capturing dairy goat morphology.
Ours
PointStack
Body Size
MAE
MAPE
MAE
MAPE
Body Length
21.9
2.81
25.7
3.24
Body Height
16.8
2.35
18.3
2.54
Chest Width
14.6
6.81
11.3
5.43
Chest Girth
15.5
1.64
28.5
3.08
Hip Height
13.1
1.77
16.2
2.16
Table 3: Comparison with improved PointStack-based body dimension measurements derived from tri-view point cloud data. Errors in mm.
Chamfer Distance
Mesh-to-Scan
mean
std
Mean
std
SMAL
89.88
9.04
68.17
7.34
SMAL+
76.16
8.34
59.32
6.07
Ours
26.58
3.17
19.66
1.74
Table 4: Comparison of monocular recovery results between ours and SMAL as well as SMAL+ models. Errors in mm.
Figure 9: Monocular 3D goat mesh recovery results. From left to right:(a) Input RGBD image; (b) SMALify with SMAL model; (c) SMALify with SMAL+ model; (d) AniMer model’s predicted mesh overlaps with the input data; (e) Our model’s predicted mesh overlaps with the input data; (f) Ground truth (GT) mesh overlaps with the RGB image.
Figure 10: MAE results for body measurements: comparison between single-view SaanenGoat reconstructions and SMAL/SMAL+ baseline models. Errors in mm.
Monocular 3D animal reconstruction is challenging due to complex articulation, self-occlusion, and fine-scale details such as fur. Existing methods often produce distorted geometry and inconsistent textures due to the lack of articulated 3D supervision and limited availability of back-view images in 2D datasets, which makes reconstructing unobserved regions particularly difficult. To address these limitations, we propose DogWeave, a model-based framework for reconstructing high-fidelity 3D canine models from a single RGB image. DogWeave improves geometry by refining a coarsely-initiated parametric mesh into a detailed SDF representation through multi-view normal field optimization using diffusion-enhanced normals. It then generates view-consistent textures through conditional partial inpainting guided by structure and style cues, enabling realistic reconstruction of unobserved regions. Using only about 7,000 dog images processed via our 2D pipeline for training, DogWeave produces complete, realistic 3D models and outperforms state-of-the-art single image to 3d reconstruction methods in both shape accuracy and texture realism for canines.
Shufan Sun, Chenchen Wang, Zongfu Yu
University of Wisconsin–Madison, Madison, WI 53706, USA
3D animal reconstruction in the wild remains challenging due to large species variation, frequent occlusions, and the prevalence of multi-animal scenes, while existing methods predominantly focus on single-animal settings. We present SAM 3D Animal, the first promptable framework for multi-animal 3D reconstruction from a single image. Built on the SMAL+ parametric animal model, our method jointly reconstructs multiple instances and supports flexible prompts in the form of keypoints and masks which enable more reliable disambiguation in crowded and occluded scenes. To train such a model, we further introduce Herd3D, a multi-animal 3D dataset containing over 5K images, designed to increase diversity in species, interactions, and occlusion patterns. Experiments on the Animal3D, APTv2, and Animal Kingdom datasets show that our framework achieves state-of-the-art results over both existing model-based and model-free methods, demonstrating a scalable and effective solution for prompt-driven animal 3D reconstruction in the wild.
Xuyi Hu, Jin Lyu, Jiuming Liu +4
University of Cambridge · Southern University of Science and Technology · Tsinghua University +1
We propose HeadsUp, a scalable feed-forward method for reconstructing high-quality 3D Gaussian heads from large-scale multi-camera setups. Our method employs an efficient encoder-decoder architecture that compresses input views into a compact latent representation. This latent representation is then decoded into a set of UV-parameterized 3D Gaussians anchored to a neutral head template. This UV representation decouples the number of 3D Gaussians from the number and resolution of input images, enabling training with many high-resolution input views. We train and evaluate our model on an internal dataset with more than 10,000 subjects, which is an order of magnitude larger than existing multi-view human head datasets. HeadsUp achieves state-of-the-art reconstruction quality and generalizes to novel identities without test-time optimization. We extensively analyze the scaling behavior of our model across identities, views, and model capacity, revealing practical insights for quality-compute trade-offs. Finally, we highlight the strength of our latent space by showcasing two downstream applications: generating novel 3D identities and animating the 3D heads with expression blendshapes.