Fashion Landmark Detection in the Wild

By Ziwei Liu, Sijie Yan, Ping Luo, Xiaogang Wang, Xiaoou Tang. 2016

173 citations recorded by the source corpus.

Visual fashion analysis has attracted many attentions in the recent years. Previous work represented clothing regions by either bounding boxes or human joints. This work presents fashion landmark detection or fashion alignment, which is to predict the positions of functional key points defined on the fashion items, such as the corners of neckline, hemline, and cuff. To encourage future studies, we introduce a fashion landmark dataset (The dataset is available at http://mmlab.ie.cuhk.edu.hk/projects/DeepFashion/LandmarkDetection.html.) with over 120K images, where each image is labeled with eight landmarks. With this dataset, we study fashion alignment by cascading multiple convolutional neural networks in three stages. These stages gradually improve the accuracies of landmark predictions. Extensive experiments demonstrate the effectiveness of the proposed method, as well as its generalization ability to pose estimation. Fashion landmark is also compared to clothing bounding boxes and human joints in two applications, fashion attribute prediction and clothes retrieval, showing that fashion landmark is a more discriminative representation to understand fashion images.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

GAN Inversion: A Survey (2021, background): “Besides, some GAN inversion studies also use other datasets [44], [45], [46], [47], [48], [49] in their experiments, such as DeepFashion [50], [51], [52], AnimeFaces [33], and StreetScapes [53].”
A Disentangling Invertible Interpretation Network for Explaining Latent Representations (2020, methodology): “The subsequent experiments use the following datasets: AnimalFaces [28], DeepFashion [29, 31], CelebA [30] MNIST [27], Cifar10 [25], and FashionMNIST [50].”
Not All Pixels Are Equal: Difficulty-Aware Semantic Segmentation via Deep Layer Cascade (2017, background): “Network cascades [2, 18, 26, 33, 24] have been studied to improve the performance in classi-ﬁcation [26], detection [18], and pose estimation [33].”
DeepFashion2: A Versatile Benchmark for Detection, Pose Estimation, Segmentation and Re-Identification of Clothing Images (2019, background): “There are various tasks that analyze clothing images such as clothes detection [2, 14], landmark prediction [15, 19, 17], clothes segmentation [18, 20, 13], and retrieval [7, 5, 14].”
TextureGAN: Controlling Deep Image Synthesis with Texture Patches (2017, methodology): “We train TextureGAN on three object-centric datasets – handbags [49], shoes [45] and clothes [27, 28, 30, 31].”
Annotating Objects and Relations in User-Generated Videos (2019, background): “In specific domains, such as the fashion and food recognition, the analytic granularity is even refined into landmarks [15] and ingredients [4], respectively, in order to boost the retrieval performance with more discriminative representation.”
Anytime Recognition with Routing Convolutional Networks (2019, methodology): “Deep learning methods that use network cascades to improve image classiﬁcation [30], object detection [4], [23], semantic segmentation [24], landmark detection [26], and pose estimation [36] are also related to our work.”
Tex2Shape: Detailed Full Human Body Geometry From a Single Image (2019, methodology): “We compare against ground truth on our own dataset and show qualitative results on 3DPW [51], DeepFashion [30, 31], and PeopleSnapshot [4] datasets.”
Towards Photo-Realistic Virtual Try-On by Adaptively Generating↔Preserving Image Content (2020, background): “Most of the existing works focus on clothing compatibility and matching learning [26, 17, 41], clothing landmark detection [30, 46, 8, 25], and fashion image analysis [14, 11, 28].”
A Comprehensive Analysis of Deep Regression (2018, unknown): “Even if some authors devote time and efforts to make their research reproducible [24], many published studies do not describe implementation, practical, or data pre-processing in detail [16], [18], [19], [22].”
Open Compound Domain Adaptation (2019, methodology): “Unlike existing curriculum adaptation methods [56, 6, 29, 25, 58, 57] that rely on some holistic measure of instance difficulty, we schedule the learning of unlabeled instances in the compound target domain according to their individual gaps to the labeled source domain, so that we solve an incremen”
A Variational U-Net for Conditional Appearance and Shape Generation (2018, methodology): “Datasets To compare with other methods, we evaluate on: shoes [43], handbags [49], Market-1501 [47], Deep-Fashion [21, 23] and COCO [20].”
Wide-Context Semantic Image Extrapolation (2019, methodology): “2) 64 × 128 → 256× 128 (used for DeepFashion); 3) 256× 256→ 256× 512 (on Paris street view, Places2, and Cityscapes).”