1 code implementation • ECCV 2020 • Christian Simon, Piotr Koniusz, Richard Nock, Mehrtash Harandi
Inspired by optimization techniques, we propose a novel meta-learning algorithm with gradient modulation to encourage fast-adaptation of neural networks in the absence of abundant data.
no code implementations • 31 Mar 2024 • Yao Ni, Piotr Koniusz
Our work addresses this gap by identifying a critical flaw in BN: the tendency for gradient explosion during the centering and scaling steps.
no code implementations • 21 Mar 2024 • Peipei Song, Jing Zhang, Piotr Koniusz, Nick Barnes
Existing eye fixation prediction methods perform the mapping from input images to the corresponding dense fixation maps generated from raw fixation points.
no code implementations • 7 Feb 2024 • Lei Wang, Jun Liu, Liang Zheng, Tom Gedeon, Piotr Koniusz
For a support sequence, we match it with view-simulated query sequences, as in the popular Dynamic Time Warping (DTW).
no code implementations • 3 Dec 2023 • Wenlong Shi, Changsheng Lu, Ming Shao, Yinjie Zhang, Siyu Xia, Piotr Koniusz
Thirdly, we propose a decoding module to include the supervision of shape masks and edges and align the original and reconstructed shape features, enforcing the learned features to be more shape-aware.
1 code implementation • 28 Oct 2023 • Maryam Haghighat, Peyman Moghadam, Shaheer Mohamed, Piotr Koniusz
In this paper, we propose an Image Modeling framework based on random orthogonal projection instead of binary masking as in MIM.
no code implementations • 27 Oct 2023 • Yifei Zhang, Hao Zhu, Jiahong Liu, Piotr Koniusz, Irwin King
We show that in the hyperbolic space one has to address the leaf- and height-level uniformity which are related to properties of trees, whereas in the ambient space of the hyperbolic manifold, these notions translate into imposing an isotropic ring density towards boundaries of Poincar\'e ball.
no code implementations • 16 Oct 2023 • Lei Wang, Piotr Koniusz
Various research studies indicate that action recognition performance highly depends on the types of motions being extracted and how accurate the human actions are represented.
no code implementations • 9 Oct 2023 • Lei Wang, Piotr Koniusz, Tom Gedeon, Liang Zheng
As such, enforcing a high similarity for positive pairs and a low similarity for negative pairs may not always be achievable, and in the case of some pairs, forcing so may be detrimental to the performance.
no code implementations • 24 Sep 2023 • Can Peng, Piotr Koniusz, Kaiyu Guo, Brian C. Lovell, Peyman Moghadam
Deep learning models suffer from catastrophic forgetting when being fine-tuned with samples of new classes.
1 code implementation • 18 Jul 2023 • Zhibin Li, Piotr Koniusz, Lu Zhang, Daniel Edward Pagendam, Peyman Moghadam
Instead of modelling statistics of features globally (i. e., by the covariance matrix of features), we learn a global field dependency matrix that captures dependencies between fields and then we refine the global field dependency matrix at the instance-wise level with different weights (so-called local dependency modelling) w. r. t.
no code implementations • CVPR 2023 • Dahyun Kang, Piotr Koniusz, Minsu Cho, Naila Murray
For this mixed setup, we propose to improve the pseudo-labels using a pseudo-label enhancer that was trained using the available ground-truth pixel-level labels.
no code implementations • 9 May 2023 • Arian Prabowo, Hao Xue, Wei Shao, Piotr Koniusz, Flora D. Salim
A road network, in the context of traffic forecasting, is typically modeled as a graph where the nodes are sensors that measure traffic metrics (such as speed) at that location.
1 code implementation • 9 May 2023 • Arian Prabowo, Hao Xue, Wei Shao, Piotr Koniusz, Flora D. Salim
During inference, the spatial encoder only requires two days of traffic data on the new roads and does not require any re-training.
1 code implementation • CVPR 2023 • Hao Zhu, Piotr Koniusz
In this paper, we propose a novel prototype-based label propagation to solve these issues.
1 code implementation • CVPR 2023 • Saimunur Rahman, Piotr Koniusz, Lei Wang, Luping Zhou, Peyman Moghadam, Changming Sun
Our work obtains a partial correlation based deep visual representation and mitigates the small sample problem often encountered by covariance matrix estimation in CNN.
no code implementations • 6 Apr 2023 • Changsheng Lu, Hao Zhu, Piotr Koniusz
Unlike current deep keypoint detectors that are trained to recognize limited number of body parts, few-shot keypoint detection (FSKD) attempts to localize any keypoints, including novel or base keypoints, depending on the reference samples.
no code implementations • CVPR 2023 • Lei Wang, Piotr Koniusz
We split action sequences into temporal blocks, Higher-order Transformer (HoT) produces embeddings of each temporal block based on (i) the body joints, (ii) pairwise links of body joints and (iii) higher-order hyper-edges of skeleton body joints.
1 code implementation • 20 Feb 2023 • Arian Prabowo, Wei Shao, Hao Xue, Piotr Koniusz, Flora D. Salim
Further analysis also shows that each pair of sensors also has a unique dynamic.
1 code implementation • 14 Feb 2023 • Hongguang Zhang, Limeng Zhang, Yuchao Dai, Hongdong Li, Piotr Koniusz
Contemporary deep learning multi-scale deblurring models suffer from many issues: 1) They perform poorly on non-uniformly blurred images/videos; 2) Simply increasing the model depth with finer-scale levels cannot improve deblurring; 3) Individual RGB frames contain a limited motion information for deblurring; 4) Previous models have a limited robustness to spatial transformations and noise.
1 code implementation • ICCV 2023 • Zhongyan Zhang, Lei Wang, Luping Zhou, Piotr Koniusz
To this end, we propose a novel feature learning framework for instance image retrieval, which embeds local spatial context information into the learned global feature representations.
no code implementations • 2 Dec 2022 • Yifei Zhang, Hao Zhu, Zixing Song, Piotr Koniusz, Irwin King
Although augmentations (e. g., perturbation of graph edges, image crops) boost the efficiency of Contrastive Learning (CL), feature level augmentation is another plausible, complementary yet not well researched strategy.
1 code implementation • 30 Oct 2022 • Shan Zhang, Naila Murray, Lei Wang, Piotr Koniusz
To address these drawbacks, we propose a Time-rEversed diffusioN tEnsor Transformer (TENET), which i) forms high-order tensor representations that capture multi-way feature occurrences that are highly discriminative, and ii) uses a transformer that dynamically extracts correlations between the query image and the entire support set, instead of a single average-pooled support embedding.
1 code implementation • 30 Oct 2022 • Lei Wang, Piotr Koniusz
Dynamic Time Warping (DTW) is used for matching pairs of sequences and celebrated in applications such as forecasting the evolution of time series, clustering time series or even matching sequence pairs in few-shot action recognition.
no code implementations • 30 Oct 2022 • Lei Wang, Piotr Koniusz
To factor out misalignment between query and support sequences of 3D body joints, we propose an advanced variant of Dynamic Time Warping which jointly models each smooth path between the query and support frames to achieve simultaneously the best alignment in the temporal and simulated camera viewpoint spaces for end-to-end learning under the limited few-shot training data.
1 code implementation • 9 Jun 2022 • Yifei Zhang, Hao Zhu, Zixing Song, Piotr Koniusz, Irwin King
In this paper, we show that the node embedding obtained via the graph augmentations is highly biased, somewhat limiting contrastive models from learning discriminative features for downstream tasks.
1 code implementation • 26 Mar 2022 • Yujiao Shi, Xin Yu, Liu Liu, Dylan Campbell, Piotr Koniusz, Hongdong Li
We address the problem of ground-to-satellite image geo-localization, that is, estimating the camera latitude, longitude and orientation (azimuth angle) by matching a query image captured at the ground level against a large-scale database with geotagged satellite images.
no code implementations • 13 Feb 2022 • Yifei Zhang, Hao Zhu, Ziqiao Meng, Piotr Koniusz, Irwin King
However, in the updating stage, all nodes share the same updating function.
1 code implementation • 15 Jan 2022 • Hongguang Zhang, Hongdong Li, Piotr Koniusz
The goal of multi-level feature design is to extract feature representations at different layer-wise levels of CNN, realizing several levels of visual abstraction to achieve robust few-shot learning.
1 code implementation • NeurIPS 2021 • Hao Zhu, Ke Sun, Piotr Koniusz
Starting from a GAN-inspired contrastive formulation, we show that the Jensen-Shannon divergence underlying many contrastive graph embedding models fails under disjoint positive and negative distributions, which may naturally emerge during sampling in the contrastive setting.
1 code implementation • CVPR 2022 • Hao Zhu, Piotr Koniusz
We present an unsupErvised discriminAnt Subspace lEarning (EASE) that improves transductive few-shot learning performance by learning a linear projection onto a subspace built from features of the support set and the unlabeled query set in the test time.
no code implementations • CVPR 2022 • Shan Zhang, Lei Wang, Naila Murray, Piotr Koniusz
We design a Kernelized Few-shot Object Detector by leveraging kernelized matrices computed over multiple proposal regions, which yield expressive non-linear representations whose model complexity is learned on the fly.
no code implementations • 23 Dec 2021 • Lei Wang, Jun Liu, Piotr Koniusz
In this paper, we propose a Few-shot Learning pipeline for 3D skeleton-based action recognition by Joint tEmporal and cAmera viewpoiNt alIgnmEnt (JEANIE).
no code implementations • CVPR 2022 • Yao Ni, Piotr Koniusz, Richard Hartley, Richard Nock
In our design, the manifold learning and coding steps are intertwined with layers of the discriminator, with the goal of attracting intermediate feature representations onto manifolds.
1 code implementation • CVPR 2022 • Changsheng Lu, Piotr Koniusz
Current non-rigid object keypoint detectors perform well on a chosen kind of species and body parts, and require a large amount of labelled keypoints for training.
no code implementations • 26 Oct 2021 • Christian Simon, Piotr Koniusz, Mehrtash Harandi
Even with the luxury of having abundant data, multi-label classification is widely known to be a challenging task to address.
no code implementations • 23 Oct 2021 • Christian Simon, Piotr Koniusz, Lars Petersson, Yan Han, Mehrtash Harandi
Our empirical evaluations show that the noise injecting operation does not degrade the performance of the NAS algorithm if the data is indeed clean.
no code implementations • 22 Oct 2021 • Yusuf Tas, Piotr Koniusz
For the image-based task, we employ the DeepFashion dataset in which we seek nearest neighbor images of positive and negative target images of the MMD data.
no code implementations • 11 Oct 2021 • Lei Wang, Ke Sun, Piotr Koniusz
We aim at capturing high-order statistics of feature vectors formed by a neural network, and propose end-to-end second- and higher-order pooling to form a tensor descriptor.
Ranked #2 on Scene Recognition on YUP++ (using extra training data)
no code implementations • 24 Aug 2021 • Hao Zhu, Piotr Koniusz
Moreover, we design a simple but efficient spectral filter for network enhancement to obtain higher-order information for node representation.
no code implementations • 24 Jul 2021 • Hao Zhu, Piotr Koniusz
Although Graph Convolutional Networks (GCNs) have demonstrated their power in various applications, the graph convolutional layers, as the most important component of GCN, are still using linear transformations and a simple pooling step.
no code implementations • 19 May 2021 • Wei Shao, Arian Prabowo, Sichen Zhao, Piotr Koniusz, Flora D. Salim
To model and forecast flight delays accurately, it is crucial to harness various vehicle trajectory and contextual sensor data on airport tarmac areas.
1 code implementation • CVPR 2021 • Christian Simon, Piotr Koniusz, Mehrtash Harandi
This in practice is done by minimizing the dissimilarity between current and previous responses of the network one way or another.
2 code implementations • ICLR 2021 • Hao Zhu, Piotr Koniusz
Our spectral analysis shows that our simple spectral graph convolution used in S^2GC is a low-pass filter which partitions networks into a few large parts.
Ranked #1 on Node Clustering on Wiki
1 code implementation • 28 Dec 2020 • Piotr Koniusz, Lei Wang, Anoop Cherian
In this paper, we propose novel tensor representations for compactly capturing such higher-order relationships between visual features for the task of action recognition.
Ranked #2 on Skeleton Based Action Recognition on UT-Kinect
Action Recognition In Videos Skeleton Based Action Recognition
no code implementations • 27 Dec 2020 • Piotr Koniusz, Hongguang Zhang
Our layer combines the feature vectors and their respective spatial locations in the feature maps produced by the last convolutional layer of CNN into a positive definite matrix with second-order statistics to which PN operators are applied, forming so-called Second-order Pooling (SOP).
no code implementations • 23 Sep 2020 • Weiwei Hou, Hanna Suominen, Piotr Koniusz, Sabrina Caldwell, Tom Gedeon
Sentence compression is a Natural Language Processing (NLP) task aimed at shortening original sentences and preserving their key information.
no code implementations • 10 Feb 2020 • Xin Yu, Zheyu Zhuang, Piotr Koniusz, Hongdong Li
In this paper, we aim to reduce such errors by incorporating the distances between pixels and keypoints into our objective.
no code implementations • 9 Feb 2020 • Xianjing Wang, Flora D. Salim, Yongli Ren, Piotr Koniusz
Point-of-Interest (POI) recommendation is one of the most important location-based services helping people discover interesting venues or services.
no code implementations • 14 Jan 2020 • Lei Wang, Piotr Koniusz
In this paper, we build on a concept of self-supervision by taking RGB frames as input to learn to predict both action concepts and auxiliary descriptors e. g., object descriptors.
Ranked #1 on Scene Recognition on YUP++ (using extra training data)
1 code implementation • ECCV 2020 • Hongguang Zhang, Li Zhang, Xiaojuan Qi, Hongdong Li, Philip H. S. Torr, Piotr Koniusz
Such encoded blocks are aggregated by permutation-invariant pooling to make our approach robust to varying action lengths and long-range temporal dependencies whose patterns are unlikely to repeat even in clips of the same class.
Ranked #6 on Few Shot Action Recognition on Kinetics-100
1 code implementation • CVPR 2021 • Hongguang Zhang, Piotr Koniusz, Songlei Jian, Hongdong Li, Philip H. S. Torr
The majority of existing few-shot learning methods describe image relations with binary labels.
no code implementations • 6 Jan 2020 • Hongguang Zhang, Philip H. S. Torr, Piotr Koniusz
In this paper, we study the impact of scale and location mismatch in the few-shot learning scenario, and propose a novel Spatially-aware Matching (SM) scheme to effectively perform matching across multiple scales and locations, and learn image relations by giving the highest weights to the best matching pairs.
no code implementations • 24 Sep 2019 • Arian Prabowo, Piotr Koniusz, Wei Shao, Flora D. Salim
This paper introduces COLTRANE, ConvolutiOnaL TRAjectory NEtwork, a novel deep map inference framework which operates on GPS trajectories collected in various environments.
1 code implementation • 24 Jun 2019 • Lei Wang, Du. Q. Huynh, Piotr Koniusz
Video-based human action recognition is currently one of the most active research areas in computer vision.
Ranked #85 on Skeleton Based Action Recognition on NTU RGB+D
no code implementations • ICCV 2019 • Lei Wang, Piotr Koniusz, Du. Q. Huynh
Thus, we propose an end-to-end trainable network with streams which learn the IDT-based BoW/FV representations at the training stage and are simple to integrate with the I3D model.
Ranked #3 on Scene Recognition on YUP++ (using extra training data)
no code implementations • ICLR 2019 • Christian Simon, Piotr Koniusz, Mehrtash Harandi
Generalization from limited examples, usually studied under the umbrella of meta-learning, equips learning techniques with the ability to adapt quickly in dynamical environments and proves to be an essential aspect of lifelong learning.
no code implementations • 7 Apr 2019 • Fatemeh Shiri, Xin Yu, Fatih Porikli, Richard Hartley, Piotr Koniusz
%Our method can recover high-quality photorealistic faces from unaligned portraits while preserving the identity of the face images as well as it can reconstruct a photorealistic face image with a desired set of attributes.
no code implementations • 7 Apr 2019 • Fatemeh Shiri, Xin Yu, Fatih Porikli, Richard Hartley, Piotr Koniusz
We develop an Identity-preserving Face Recovery from Portraits (IFRP) method that utilizes a Style Removal network (SRN) and a Discriminative Network (DN).
1 code implementation • CVPR 2019 • Hongguang Zhang, Yuchao Dai, Hongdong Li, Piotr Koniusz
depth, we propose a stacked version of our multi-patch model.
Ranked #9 on Deblurring on RealBlur-R (trained on GoPro) (SSIM (sRGB) metric)
no code implementations • CVPR 2019 • Hongguang Zhang, Jing Zhang, Piotr Koniusz
To the best of our knowledge, we are the first to leverage saliency maps for such a task and we demonstrate their usefulness in hallucinating additional datapoints for few-shot learning.
1 code implementation • 11 Mar 2019 • Ke Sun, Piotr Koniusz, Zhen Wang
We try to minimize the loss wrt the perturbed $G+\Delta{G}$ while making $\Delta{G}$ to be effective in terms of the Fisher information of the neural network.
no code implementations • 10 Nov 2018 • Hongguang Zhang, Piotr Koniusz
In this paper, we propose a similarity learning network leveraging second-order information and Power Normalizations.
no code implementations • 8 Nov 2018 • Hongguang Zhang, Piotr Koniusz
Specifically, we leverage two sources of datapoints (observed and auxiliary) to train some classifier to recognize which test datapoints come from seen and which from unseen classes.
Generalized Zero-Shot Learning Generative Adversarial Network +1
no code implementations • ECCV 2018 • Piotr Koniusz, Yusuf Tas, Hongguang Zhang, Mehrtash Harandi, Fatih Porikli, Rui Zhang
To achieve robust baselines, we build on a recent approach that aligns per-class scatter matrices of the source and target CNN streams.
no code implementations • ECCV 2018 • Tsung-Yu Lin, Subhransu Maji, Piotr Koniusz
In this paper, we study a class of orderless aggregation functions designed to minimize interference or equalize contributions in the context of second-order features and we show that they can be computed just as efficiently as their first-order counterparts and they have favorable properties over aggregation by summation.
no code implementations • CVPR 2018 • Piotr Koniusz, Hongguang Zhang, Fatih Porikli
In this paper, we reconsider these operators in the deep learning setup by introducing a novel layer that implements PN for non-linear pooling of feature maps.
no code implementations • 24 Jun 2018 • Yusuf Tas, Piotr Koniusz
In this paper, we propose a new representation which encodes sequences of 3D body skeleton joints in texture-like representations derived from mathematically rigorous kernel methods.
no code implementations • 24 Jun 2018 • Rui Zhang, Yusuf Tas, Piotr Koniusz
We discuss the application of wearable cameras, and the practical and technical challenges in devising a robust system that can recognize artworks viewed by the visitors to create a detailed record of their visit.
no code implementations • CVPR 2018 • Hongguang Zhang, Piotr Koniusz
In contrast, we apply well-established kernel methods to learn a non-linear mapping between the feature and attribute spaces.
no code implementations • 5 Feb 2018 • Fatemeh Shiri, Xin Yu, Fatih Porikli, Piotr Koniusz
To enforce the destylized faces to be similar to authentic face images, we employ a discriminative network, which consists of convolutional and fully connected layers.
no code implementations • 4 Feb 2018 • Piotr Koniusz, Yusuf Tas, Hongguang Zhang, Mehrtash Harandi, Fatih Porikli, Rui Zhang
To achieve robust baselines, we build on a recent approach that aligns per-class scatter matrices of the source and target CNN streams [15].
no code implementations • 8 Jan 2018 • Fatemeh Shiri, Xin Yu, Fatih Porikli, Richard Hartley, Piotr Koniusz
In this paper, we present a new Identity-preserving Face Recovery from Portraits (IFRP) to recover latent photorealistic faces from unaligned stylized portraits.
no code implementations • 19 Jan 2017 • Anoop Cherian, Piotr Koniusz, Stephen Gould
The HOK descriptors are then generated from the higher-order co-occurrences of these feature maps, and are then used as input to a video-level classifier.
no code implementations • CVPR 2017 • Piotr Koniusz, Yusuf Tas, Fatih Porikli
In this paper, we propose an approach to the domain adaptation, dubbed Second- or Higher-order Transfer of Knowledge (So-HoT), based on the mixture of alignments of second- or higher-order scatter statistics between the source and target domains.
no code implementations • CVPR 2016 • Piotr Koniusz, Anoop Cherian
Super-symmetric tensors - a higher-order extension of scatter matrices - are becoming increasingly popular in machine learning and computer vision for modeling data statistics, co-occurrences, or even as visual descriptors.
no code implementations • 1 Apr 2016 • Piotr Koniusz, Anoop Cherian, Fatih Porikli
We first define RBF kernels on 3D joint sequences, which are then linearized to form kernel descriptors.
no code implementations • 9 Sep 2015 • Piotr Koniusz, Anoop Cherian
Super-symmetric tensors - a higher-order extension of scatter matrices - are becoming increasingly popular in machine learning and computer vision for modelling data statistics, co-occurrences, or even as visual descriptors.
no code implementations • NeurIPS 2014 • Julien Mairal, Piotr Koniusz, Zaid Harchaoui, Cordelia Schmid
An important goal in visual recognition is to devise image representations that are invariant to particular transformations.
Ranked #24 on Image Classification on MNIST