Patents by Inventor Ruimao Zhang
Ruimao Zhang has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20220415007Abstract: Methods, systems, electronic devices, and computer-readable storage media for image normalization processing are provided. In one aspect, an image normalization processing method includes: normalizing a feature map by respectively using K normalization factors to obtain K candidate normalized feature maps; for each of the K normalization factors, determining a first weight value for the normalization factor; and determining a target normalized feature map corresponding to the feature map based on the candidate normalized feature map corresponding to each of the K normalization factors and the first weight value for each of the K normalization factors. The K candidate normalized feature maps and the K normalization factors have a one-to-one correspondence, and K is an integer greater than 1.Type: ApplicationFiled: August 23, 2022Publication date: December 29, 2022Inventors: Ruimao ZHANG, Zhanglin PENG, Lingyun WU, Ping LUO
-
Patent number: 11430167Abstract: Methods, systems, apparatus, and non-transitory computer readable storage media for image processing are provided. In one aspect, an image processing method includes: generating a garment deformation template image and a first human body template image based on a first semantic segmentation image of a human body in a first image and a target garment image, generating a target garment deformation image by performing deformation on the target garment image based on the garment deformation template image, obtaining a second human body template image by adjusting the first human body template image based on a second semantic segmentation image of the human body in the first image and the garment deformation template image, and transforming the first image into a second image including the human body wearing a target garment based on the target garment deformation image, the second human body template image, and the garment deformation template image.Type: GrantFiled: November 24, 2021Date of Patent: August 30, 2022Assignee: SHENZHEN SENSETIME TECHNOLOGY CO., LTD.Inventors: Ruimao Zhang, Ping Luo
-
Publication number: 20220084270Abstract: Methods, systems, apparatus, and non-transitory computer readable storage media for image processing are provided. In one aspect, an image processing method includes: generating a garment deformation template image and a first human body template image based on a first semantic segmentation image of a human body in a first image and a target garment image, generating a target garment deformation image by performing deformation on the target garment image based on the garment deformation template image, obtaining a second human body template image by adjusting the first human body template image based on a second semantic segmentation image of the human body in the first image and the garment deformation template image, and transforming the first image into a second image including the human body wearing a target garment based on the target garment deformation image, the second human body template image, and the garment deformation template image.Type: ApplicationFiled: November 24, 2021Publication date: March 17, 2022Inventors: Ruimao ZHANG, Ping LUO
-
Patent number: 11222231Abstract: Target matching method and apparatus, electronic device, and storage medium, including: extracting feature vector of each frame in query image sequence and feature vector of each frame in candidate image sequence; determining self-expression feature vector of query image sequence, collaborative expression feature vector of the query image sequence, self-expression feature vector of candidate image sequence, and collaborative expression feature vector of candidate image sequence based on feature vector of each frame in query image sequence and feature vector of each frame in candidate image sequence; determining similarity feature vector between query image sequence and candidate image sequence based on self-expression feature vector of query image sequence, collaborative expression feature vector of query image sequence, self-expression feature vector of candidate image sequence, and collaborative expression feature vector of candidate image sequence; and determining matching result between query image sequenType: GrantFiled: April 7, 2020Date of Patent: January 11, 2022Assignee: SHENZHEN SENSETIME TECHNOLOGY CO., LTD.Inventors: Ruimao Zhang, Hongbin Sun, Ping Luo, Yuying Ge, Kuanze Ren, Liang Lin, Xiaogang Wang
-
Publication number: 20210312289Abstract: The present disclosure relates to a data processing method and apparatus, and a storage medium. The method includes: inputting input data into a neural network model to obtain feature data currently output by a network layer in the neural network model (S100); determining, according to transformation parameters of the neural network model, a normalization mode matched with the feature data (S200), wherein the transformation parameters are used for adjusting a statistical range of statistics of the feature data, and the statistical range is used for representing the normalization mode; and performing normalization processing on the feature data according to the determined normalization mode to obtain normalized feature data (S300). According to embodiments of the present disclosure, the purpose of autonomously learning a matched normalization mode for each normalization layer of the neural network model can be implemented without human intervention.Type: ApplicationFiled: June 18, 2021Publication date: October 7, 2021Applicant: SHENZHEN SENSETIME TECHNOLOGY CO., LTD.Inventors: Ping LUO, Lingyun WU, Zhanglin PENG, Ruimao ZHANG, Jiamin REN, Wenqi SHAO
-
Publication number: 20210287091Abstract: A neural network training method and apparatus and an image matching method and apparatus are provided. The neural network training method at least includes: labeling annotation information of a first clothing instance and a second clothing instance, where the first clothing instance and the second clothing instance are respectively from a first clothing image and a second clothing image; pairing the first clothing image and the second clothing image in response to a state of matching between the first clothing instance and the second clothing instance; and training a neural network to be trained based on the paired first clothing image and second clothing image.Type: ApplicationFiled: June 2, 2021Publication date: September 16, 2021Applicant: SHENZHEN SENSETIME TECHNOLOGY CO., LTD.Inventors: Yuying GE, Lingyun WU, Ruimao ZHANG, Ping LUO
-
Publication number: 20210158088Abstract: An image processing method and apparatus, and a storage medium are provided. The method includes: obtaining a first feature map of an image to be processed (S101); determining a final weight vector of the first feature map (S102); determining a target normalization mode corresponding to the first feature map from a preset normalization set is determined according to the final weight vector (S103); and normalizing the first feature map by means of the target normalization mode to obtain a second feature map (S104).Type: ApplicationFiled: February 8, 2021Publication date: May 27, 2021Inventors: Wenqi Shao, Tianjian Meng, Ruimao Zhang, Ping Luo, Lingyun Wu
-
Patent number: 10776685Abstract: This invention is an image retrieval method based on bit-scalable deep hashing learning. According to the method, the training images is used to generate a batch of image triples, wherein each of the triples contains two images with the same label and one image with a different label. The purpose of model training is to maximize a margin between matched image pairs and unmatched image pairs in the Hamming space. The deep convolutional neural network is utilized to train the model in an end-to-end fasion, where discriminative images features and has functions are simultaneously optimized. Furthermore, each bit of the hashing codes is unequally weighted so that we can manipulate the code length by truncating the insignificant bits. It is also shown that the generated bit-scalable hashing codes well preserve the discriminative powers with sorter code lengths.Type: GrantFiled: May 25, 2018Date of Patent: September 15, 2020Assignee: SUN YAT-SEN UNIVERSITYInventors: Liang Lin, Ruimao Zhang, Qing Wang, Bo Jiang
-
Publication number: 20200257979Abstract: Embodiments of the present disclosure disclose normalization methods and apparatuses for a deep neural network, devices, and storage media. The method includes: inputting an input data set into a deep neural network, the input data set including at least one piece of input data; normalizing a feature map set output by means of a network layer in the deep neural network from at least one dimension to obtain at least one dimension variance and at least one dimension mean; and determining a normalized target feature map set based on the at least one dimension variance and the at least one dimension mean. Based on the embodiments of the present disclosure, normalization is performed along at least one dimension so that statistics information of each dimension of a normalization operation is covered, thereby ensuring good robustness of statistics in each dimension without excessively depending on the batch size.Type: ApplicationFiled: April 29, 2020Publication date: August 13, 2020Inventors: Ping LUO, Lingyun Wu, Jiamin Ren, Zhanglin Peng, Ruimao Zhang, Xinjiang Wang
-
Publication number: 20200234078Abstract: Target matching method and apparatus, electronic device, and storage medium, including: extracting feature vector of each frame in query image sequence and feature vector of each frame in candidate image sequence; determining self-expression feature vector of query image sequence, collaborative expression feature vector of the query image sequence, self-expression feature vector of candidate image sequence, and collaborative expression feature vector of candidate image sequence based on feature vector of each frame in query image sequence and feature vector of each frame in candidate image sequence; determining similarity feature vector between query image sequence and candidate image sequence based on self-expression feature vector of query image sequence, collaborative expression feature vector of query image sequence, self-expression feature vector of candidate image sequence, and collaborative expression feature vector of candidate image sequence; and determining matching result between query image sequenType: ApplicationFiled: April 7, 2020Publication date: July 23, 2020Inventors: Ruimao Zhang, Hongbin Sun, Ping Luo, Yuying Ge, Kuanze Ren, Liang Lin, Xiaogang Wang
-
Publication number: 20180276528Abstract: This invention is an image retrieval method based on bit-scalable deep hashing learning. According to the method, the training images is used to generate a batch of image triples, wherein each of the triples contains two images with the same label and one image with a different label. The purpose of model training is to maximize a margin between matched image pairs and unmatched image pairs in the Hamming space. The deep convolutional neural network is utilized to train the model in an end-to-end fasion, where discriminative images features and has functions are simultaneously optimized. Furthermore, each bit of the hashing codes is unequally weighted so that we can manipulate the code length by truncating the insignificant bits. It is also shown that the generated bit-scalable hashing codes well preserve the discriminative powers with sorter code lengths.Type: ApplicationFiled: May 25, 2018Publication date: September 27, 2018Inventors: Liang Lin, Ruimao Zhang, Qing Wang, Bo Jiang