clothes classification, attribute prediction, clothing item retrieval.

• clothes have large variations in style, texture, and cutting.
• clothing items are frequently subject to deformation and occlusion.
• clothes images often exhibit serous variations when they are taken under different scenarios.

Liu, Ziwei, et al. “Deepfashion: Powering robust clothes recognition and retrieval with rich annotations.” Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.

---

Paper: Deepfashion

Summary

• introduce DeepFashion, a large-scale clothes dataset with comprehensive annotations.

• propose FashionNet,which learns clothing features by jointly predicting clothing attributes and landmarks. the landmarks are human labeled:

• FashionNet work pipeline: can be used to predict the landmark of clothes.

Relative

• (a) Additional landmark locations improve clothes recognition. (b) Massive attributes lead to better partition of the clothing feature space.

• Relative datasets

• clothing recognition and retrieval: hand-craft features like sift, hog, color histogram.

Yamaguchi, Kota, et al. “Parsing clothing in fashion photographs.” 2012 IEEE Conference on Computer vision and pattern recognition. IEEE, 2012.

---

Paper: ParsingClothing

Summary

• a novel dataset for studying clothing parsing, consisting of 158.235 fashion photos with associated text annotations, and web-based tools for labeling.

• A methods to parse pictures of people into their constituent garments. And the clothes pipelines are as follows:

  • (a) Parsing the image into Superpixels [1],

  • (b) Original pose estimation using state of the art flexible mixtures of parts model [27].

  • (c) Precise clothing parse output by our proposed clothing estimation model (note the accurate labeling of items as small as the wearer’s necklace, or as intricate as her open toed shoes).

  • (d) Optional reestimate of pose using clothing estimates (note the improvement in her left arm prediction, compared to the original incorrect estimate down along the side of her body).

  

• Prototype garment search application results. Query photo (left column) retrieves similar clothing items (right columns) independent of pose and with high visual similarity.

• dataset create funciton:

Scenarios suitable

• pose estimation: incorporated mixtures of parts to obtain state of the art results, and extending the approach to incorporate clothing estimations in models for pose identification.

Yu, Tao, et al. “Simulcap: Single-view human performance capture with cloth simulation.” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019.

---

Paper: Simulcap

Summary

• proposes a new method for live free viewpoint human performance capture with dynamic details(eg. cloth wrinkles) using single RGBD, simulate plausible cloth dynamics and cloth-body interactions even in the occluded regions.
  • can predict the occluded cloth part more accurately than the commonly used surface skinning, and non-rigid warping.
  • the observed cloth details, suh as wrinkles, can be reconstructed by formulating data fitting as a physical process.

• Cloth simulation: 这部分没有看，里面有

Patel, Chaitanya, Zhouyingcheng Liao, and Gerard Pons-Moll. “Tailornet: Predicting clothing in 3d as a function of human pose, shape and garment style.” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.

---

Paper: Tailornet

Summary

• present TailorNet, a neural model which predicts clothing deformation in 3D as a function of three factors: pose, shape and style (garment geometry), while retaining wrinkle detail.
• The first joint model of clothing style, pose and shape variation, which is simple, easy to deploy and fully differentiable for easy integration with deep learning.
• decomposition of mesh deformations into low and high-frequency components, which coupled with a mixture model, allows to retain high-frequency wrinkles.
• Pipelines: Overview of our model to predict the draped garment with style γ on the body with pose θ and shape β.
  • Low frequency of the deformations are predicted using a single model.
  • High frequency of pose dependent deformations for K prototype shape-style pairs are separately computed and mixed using a RBF kernel to get the final high frequency of the deformations.
  • The low and high frequency predictions are added to get the unposed garment output, which is posed to using standard skinning to get the garment

Relative

• animation of clothing
  • Physics Based Simulation (PBS):
  • Data-driven cloth models:
  • Pixel based models:

Chance, Greg, et al. ““elbows out”—predictive tracking of partially occluded pose for robot-assisted dressing.” IEEE Robotics and Automation Letters 3.4 (2018): 3598-3605.

---

Paper: elbows out

Summary

Hsiao, Wei-Lin, and Kristen Grauman. “ViBE: Dressing for diverse body shapes.” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.

---

Paper: ViBE

Summary

• introduce ViBE, a Visual Body-aware Embedding that captures clothing’s affinity with different body shapes,given an image of a person, the proposed embedding identifies garments that will flatter her specific body shape.

• Example categories of body shapes, with styling tips and recommended dresses for each, according to fashion blogs

Relative

• Trained largely from images of slender fashionistas and celebrities (bottom row), existing methods ignore body shape’s effect on clothing recommendation and exclude much of the spectrum of real body shapes.
• Our proposed embedding considers diverse body shapes (top row) and learns which garments flatter which across the spectrum of the real population. address the influence of body shape on garment compatibility or style.

• Fashion Styles and compatibility:
  • recognition problems, like matching items seen on the street to a catalog searching for products, or parsing an outfit into garments.
  • style-meta-patterns in what people wear with visual attri.
• Virtual try on clothing retargeting:
  • estimate garment draping on a 3D images scan;
  • retarget styles for people in 2D images or video, or render a virtual try-on with sophisticated image generation.
• Body and garment shape estimation:
  • estimating people and clothing’s 3D geometry from 2D RGB images;
• Sizing clothing: given a product and the purchase history of a user, these methods predict whether a given size will be too large, small or just right.