This paper presents the 1st place solution for the Google Universal Images Embedding Competition on Kaggle. The highlighted part of our solution is based on 1) A novel way to conduct training and fine-tuning; 2) The idea of a better ensemble in the pool of models that make embedding; 3) The potential trade-off between fine-tuning on high-resolution and overlapping patches; 4) The potential factors to work for the dynamic margin. Our solution reaches 0.728 in the private leader board, which achieve 1st place in Google Universal Images Embedding Competition.
Kawasaki Heavy Industries, Ltd.（川崎重工业株式会社）；Ferix Inc.（Ferix公司）
This paper presents the 3rd place solution to the Google Universal Image Embedding Competition on Kaggle. We use ViT-H/14 from OpenCLIP for the backbone of ArcFace, and trained in 2 stage. 1st stage is done with freezed backbone, and 2nd stage is whole model training. We achieve 0.692 mean Precision @5 on private leaderboard. Code available at https://github.com/YasumasaNamba/google-universal-image-embedding
This paper presents the 1st place solution to the Google Landmark Retrieval 2020 Competition on Kaggle. The solution is based on metric learning to classify numerous landmark classes, and uses transfer learning with two train datasets, fine-tuning on bigger images, adjusting loss weight for cleaner samples, and esemble to enhance the model's performance further. Finally, it scored 0.38677 mAP@100 on the private leaderboard.
This paper presents the 2nd place solution to the Google Landmark Retrieval 2021 Competition on Kaggle. The solution is based on a baseline with training tricks from person re-identification, a continent-aware sampling strategy is presented to select training images according to their country tags and a Landmark-Country aware reranking is proposed for the retrieval task. With these contributions, we achieve 0.52995 mAP@100 on private leaderboard. Code available at https://github.com/WesleyZhang1991/Google_Landmark_Retrieval_2021_2nd_Place_Solution
In this paper, we show our solution to the Google Landmark Recognition 2021 Competition. Firstly, embeddings of images are extracted via various architectures (i.e. CNN-, Transformer- and hybrid-based), which are optimized by ArcFace loss. Then we apply an efficient pipeline to re-rank predictions by adjusting the retrieval score with classification logits and non-landmark distractors. Finally, the ensembled model scores 0.489 on the private leaderboard, achieving the 3rd place in the 2021 edition of the Google Landmark Recognition Competition.
This paper presents the 2nd place solution to the Google Landmark Retrieval Competition 2020. We propose a training method of global feature model for landmark retrieval without post-processing, such as local feature and spatial verification. There are two parts in our retrieval method in this competition. This training scheme mainly includes training by increasing margin value of arcmargin loss and increasing image resolution step by step. Models are trained by PaddlePaddle framework and Pytorch framework, and then converted to tensorflow 2.2. Using this method, we got a public score of 0.401…
In this paper, we present our solution, which placed 5th in the kaggle Google Universal Image Embedding Competition in 2022. We use the ViT-H visual encoder of CLIP from the openclip repository as a backbone and train a head model composed of BatchNormalization and Linear layers using ArcFace. The dataset used was a subset of products10K, GLDv2, GPR1200, and Food101. And applying TTA for part of images also improves the score. With this method, we achieve a score of 0.684 on the public and 0.688 on the private leaderboard. Our code is available. https://github.com/riron1206/kaggle-Google-Unive…
LINE Corp（LINE公司）；NVIDIA（英伟达）；ReadSense Ltd（阅感有限公司）；Carnegie Mellon University（卡内基梅隆大学）
We present our third place solution to the Google Landmark Recognition 2020 competition. It is an ensemble of global features only Sub-center ArcFace models. We introduce dynamic margins for ArcFace loss, a family of tune-able margin functions of class size, designed to deal with the extreme imbalance in GLDv2 dataset. Progressive finetuning and careful postprocessing are also key to the solution. Our two submissions scored 0.6344 and 0.6289 on private leaderboard, both ranking third place out of 736 teams.
Beijing University of Posts and Telecommunications（北京邮电大学）；WeChat AI（微信人工智能）
Image retrieval is a fundamental problem in computer vision. This paper presents our 3rd place detailed solution to the Google Landmark Retrieval 2020 challenge. We focus on the exploration of data cleaning and models with metric learning. We use a data cleaning strategy based on embedding clustering. Besides, we employ a data augmentation method called Corner-Cutmix, which improves the model's ability to recognize multi-scale and occluded landmark images. We show in detail the ablation experiments and results of our method.
We present our solutions to the Google Landmark Challenges 2021, for both the retrieval and the recognition tracks. Both solutions are ensembles of transformers and ConvNet models based on Sub-center ArcFace with dynamic margins. Since the two tracks share the same training data, we used the same pipeline and training approach, but with different model selections for the ensemble and different post-processing. The key improvement over last year is newer state-of-the-art vision architectures, especially transformers which significantly outperform ConvNets for the retrieval task. We finished thi…
We present an object detection framework based on PaddlePaddle. We put all the strategies together (multi-scale training, FPN, Cascade, Dcnv2, Non-local, libra loss) based on ResNet200-vd backbone. Our model score on public leaderboard comes to 0.6269 with single scale test. We proposed a new voting method called top-k voting-nms, based on the SoftNMS detection results. The voting method helps us merge all the models' results more easily and achieve 2nd place in the Google AI Open Images Object Detection Track 2019.
As Transformer-based architectures have recently shown encouraging progresses in computer vision. In this work, we present the solution to the Google Landmark Recognition 2021 Challenge held on Kaggle, which is an improvement on our last year's solution by changing three designs, including (1) Using Swin and CSWin as backbone for feature extraction, (2) Train on full GLDv2, and (3) Using full GLDv2 images as index image set for kNN search. With these modifications, our solution significantly improves last year solution on this year competition. Our full pipeline, after ensembling Swin, CSWin, …
Chongqing University of Technology（重庆理工大学）；Institute of Intelligent Machines, Chinese Academy of Sciences（中国科学院合肥智能机械研究所）；University of Science and Technology of China（中国科学技术大学）
Image representations are a critical building block of computer vision applications. This paper presents the 2nd place solution to the Google Universal Image Embedding Competition, which is part of the ECCV2022 instance-level recognition workshops. We use the instance-level fine-grained image classification method to complete this competition. We focus on data building and processing, model structure, and training strategies. Finally, the solution scored 0.713 on the public leaderboard and 0.709 on the private leaderboard.
Democritus University of Thrace（色雷斯德谟克利特大学）；Neapolis University Pafos（帕福斯尼阿波利斯大学）
This paper presents the 6th place solution to the Google Universal Image Embedding competition on Kaggle. Our approach is based on the CLIP architecture, a powerful pre-trained model used to learn visual representation from natural language supervision. We also utilized the SubCenter ArcFace loss with dynamic margins to improve the distinctive power of class separability and embeddings. Finally, a diverse dataset has been created based on the test's set categories and the leaderboard's feedback. By carefully crafting a training scheme to enhance transfer learning, our submission scored 0.685 o…
We present a retrieval based system for landmark retrieval and recognition challenge.There are five parts in retrieval competition system, including feature extraction and matching to get candidates queue; database augmentation and query extension searching; reranking from recognition results and local feature matching. In recognition challenge including: landmark and non-landmark recognition, multiple recognition results voting and reranking using combination of recognition and retrieval results. All of models trained and predicted by PaddlePaddle framework. Using our method, we achieved 2nd …
We present a large-scale object detection system by team PFDet. Our system enables training with huge datasets using 512 GPUs, handles sparsely verified classes, and massive class imbalance. Using our method, we achieved 2nd place in the Google AI Open Images Object Detection Track 2018 on Kaggle.
University of Tsukuba（筑波大学）；Preferred Networks, Inc.（优选网络公司）；Waseda University（早稻田大学）
We propose an efficient pipeline for large-scale landmark image retrieval that addresses the diversity of the dataset through two-stage discriminative re-ranking. Our approach is based on embedding the images in a feature-space using a convolutional neural network trained with a cosine softmax loss. Due to the variance of the images, which include extreme viewpoint changes such as having to retrieve images of the exterior of a landmark from images of the interior, this is very challenging for approaches based exclusively on visual similarity. Our proposed re-ranking approach improves the resul…
Beijing Institute of Technology（北京理工大学）；University of Chinese Academy of Sciences（中国科学院大学）；Beihang University（北京航空航天大学）；Nanyang Technological University（南洋理工大学）；Peking University（北京大学）
This report details our solution to the Google AI Open Images Challenge 2019 Object Detection Track. Based on our detailed analysis on the Open Images dataset, it is found that there are four typical features: large-scale, hierarchical tag system, severe annotation incompleteness and data imbalance. Considering these characteristics, many strategies are employed, including larger backbone, distributed softmax loss, class-aware sampling, expert model, and heavier classifier. In virtue of these effective strategies, our best single model could achieve a mAP of 61.90. After ensemble, the final mA…
In this study, we propose a novel scene descriptor for visual place recognition. Unlike popular bag-of-words scene descriptors which rely on a library of vector quantized visual features, our proposed descriptor is based on a library of raw image data, such as publicly available photo collections from Google StreetView and Flickr. The library images need not to be associated with spatial information regarding the viewpoint and orientation of the scene. As a result, these images are cheaper than the database images; in addition, they are readily available. Our proposed descriptor directly mines…
3rd Place Solution for VisDA 2021 Challenge -- Universally Domain Adaptive Image Recognition
Beijing University of Posts and Telecommunications（北京邮电大学）；Tianjin University（天津大学）；Columbia University（哥伦比亚大学）；UC Berkeley（加州大学伯克利分校）；Didi Chuxing（滴滴出行）
Australian National University（澳大利亚国立大学）；Australian Centre for Robotic Vision（澳大利亚机器人视觉中心）；Northwestern Polytechnical University（西北工业大学）
This paper tackles the problem of large-scale image-based localization (IBL) where the spatial location of a query image is determined by finding out the most similar reference images in a large database. For solving this problem, a critical task is to learn discriminative image representation that captures informative information relevant for localization. We propose a novel representation learning method having higher location-discriminating power. It provides the following contributions: 1) we represent a place (location) as a set of exemplar images depicting the same landmarks and aim to m…
INRIA（法国国家信息与自动化研究所）；Tokyo Tech（东京工业大学）；CTU in Prague（布拉格捷克理工大学）
We tackle the problem of large scale visual place recognition, where the task is to quickly and accurately recognize the location of a given query photograph. We present the following three principal contributions. First, we develop a convolutional neural network (CNN) architecture that is trainable in an end-to-end manner directly for the place recognition task. The main component of this architecture, NetVLAD, is a new generalized VLAD layer, inspired by the "Vector of Locally Aggregated Descriptors" image representation commonly used in image retrieval. The layer is readily pluggable into a…
Employing Universal Voting Schemes for Improved Visual Place Recognition Performance
School of Computer Science and Electronic Engineering, University of Essex（埃塞克斯大学计算机科学与电子工程学院）；School of Electrical Engineering and Computer Science, Queensland University of Technology（昆士兰科技大学电气工程与计算机科学学院）
How to find places that are not indexed by Google Maps? We propose an intuitive method and framework to locate places based on their distinctive spatial features. The method uses satellite and street view images in machine vision approaches to classify locations. If we can classify locations, we just need to repeat for non-overlapping locations in our area of interest. We assess the proposed system in finding Parkour spots in the campus of Arizona State University. The results are very satisfactory, having found more than 25 new Parkour spots, with a rate of true positives above 60%.
University of Rochester（罗切斯特大学）；Peking University（北京大学）；Tencent Inc.（腾讯公司）
Extracting effective deep features to represent content and style information is the key to universal style transfer. Most existing algorithms use VGG19 as the feature extractor, which incurs a high computational cost and impedes real-time style transfer on high-resolution images. In this work, we propose a lightweight alternative architecture - ArtNet, which is based on GoogLeNet, and later pruned by a novel channel pruning method named Zero-channel Pruning specially designed for style transfer approaches. Besides, we propose a theoretically sound sandwich swap transform (S2) module to transf…
United States Department of Agriculture-Agricultural Research Service（美国农业部农业研究局）；Washington State University（华盛顿州立大学）；University of California-Davis（加州大学戴维斯分校）；California Polytechnic and State University（加州州立理工大学）；San Francisco State University（旧金山州立大学）；Lawrence Berkeley National Laboratory（劳伦斯伯克利国家实验室）
X-ray micro-computed tomography (X-ray microCT) has enabled the characterization of the properties and processes that take place in plants and soils at the micron scale. Despite the widespread use of this advanced technique, major limitations in both hardware and software limit the speed and accuracy of image processing and data analysis. Recent advances in machine learning, specifically the application of convolutional neural networks to image analysis, have enabled rapid and accurate segmentation of image data. Yet, challenges remain in applying convolutional neural networks to the analysis …
The Google Universal Image Embedding (GUIE) Challenge is one of the first competitions in multi-domain image representations in the wild, covering a wide distribution of objects: landmarks, artwork, food, etc. This is a fundamental computer vision problem with notable applications in image retrieval, search engines and e-commerce. In this work, we explain our 4th place solution to the GUIE Challenge, and our "bag of tricks" to fine-tune zero-shot Vision Transformers (ViT) pre-trained using CLIP.
Centro de Informatica, Universidade Federal de Pernambuco（伯南布哥联邦大学信息中心）
Many works in the recent literature introduce semantic mapping methods that use CNNs (Convolutional Neural Networks) to recognize semantic properties in images. The types of properties (eg.: room size, place category, and objects) and their classes (eg.: kitchen and bathroom, for place category) are usually predefined and restricted to a specific task. Thus, all the visual data acquired and processed during the construction of the maps are lost and only the recognized semantic properties remain on the maps. In contrast, this work introduces a topological semantic mapping method that uses deep …
Universal adversarial perturbations (UAPs), a.k.a. input-agnostic perturbations, has been proved to exist and be able to fool cutting-edge deep learning models on most of the data samples. Existing UAP methods mainly focus on attacking image classification models. Nevertheless, little attention has been paid to attacking image retrieval systems. In this paper, we make the first attempt in attacking image retrieval systems. Concretely, image retrieval attack is to make the retrieval system return irrelevant images to the query at the top ranking list. It plays an important role to corrupt the n…
