Italy


This paper studies the problem of Person Re-Identification (ReID)
for large-scale applications. Recent research efforts have been devoted
to building complicated part models, which introduce considerably
high computational cost and memory consumption, inhibiting its practicability
in large-scale applications. This paper aims to develop a novel learning
strategy to find efficient feature embeddings while maintaining the
balance of accuracy and model complexity. More specifically, we find
by enhancing the classical triplet loss together with cross-entropy
loss, our method can explore the hard examples and build a discriminant
feature embedding yet compact enough for large-scale applications.
Our method is carried out progressively using Bayesian optimization,
and we call it the Progressive Learning Algorithm (PLA). Extensive
experiments on three large-scale datasets show that our PLA is comparable
or better than the state-of-the-arts. Especially, on the challenging
Market-1501 dataset, we achieve Rank-1=94.7%/mAP=89.4% while saving
at least 30\% parameters than strong part models.

ICPR 2020

MixTConv: Mixed Temporal Convolutional Kernels for Efficient Action Recognition

action recognition

deep neural network

computer vision

ICPR 2020 is the premier international conference on Pattern Recognition, gathering leading researchers from around the world to present their latest achievements in Pattern Analysis and Recognition, applied Artificial Intelligence and Machine Learning. The topic of the year is " Putting Artificial intelligence to work on Patterns". 
For the first time, ICPR has employed a two-round review process similar to journal submissions, with new papers allowed to be submitted in either the first or second round and papers submitted in the first round and not accepted allowed to be revised and re-submitted for second round review. 

In this edition of the **ICPR 2020** Conference, there are 1407 Oral/Poster presentations organized in five Tracks: 
- Artificial Intelligence, Machine Learning for Pattern Analysis; 
- Biometrics, Human Analysis and Behavior Understanding; Computer Vision, Robotics and Intelligent Systems; 
- Document and Media Analysis; and 
- Image and Signal Processing. 

The **ICPR 2020** Main Conference program also includes 7 Keynotes from prominent world-renowned scientists, 8 Tutorials, 28 Demos, 8 Challenges, 2 Panels, as well as an Industry/Sponsor Exhibition. In addition, 41 Workshops on a range of important current and emerging research topics complete the ICPR 2020 program. 


poster

Face recognition is a widely accepted biometric verification tool, as the face contains a lot of information about the identity of a person. In this study, a 2-step neural-based pipeline is presented for matching 3D facial shape to multiple DNA-related properties (sex, age, BMI and genomic background). The first step consists of a triplet loss-based metric learner that compresses facial shape into a lower dimensional embedding while preserving information about the property of interest. Most studies in the field of metric learning have only focused on 2D Euclidean data. In this work, geometric deep learning is employed to learn directly from 3D facial meshes. To this end, spiral convolutions are used along with a novel mesh-sampling scheme that retains uniformly sampled 3D points at different levels of resolution. The second step is a multi-biometric fusion by a fully connected neural network. The network takes an ensemble of embeddings and property labels as input and returns genuine and imposter scores. Since embeddings are accepted as an input, there is no need to train classifiers for the different properties and available data can be used more efficiently. Results obtained by a 10-fold cross-validation for biometric verification show that combining multiple properties leads to stronger biometric systems. Furthermore, the proposed neural-based pipeline outperforms a linear baseline, which consists of principal component analysis, followed by classification with linear support vector machines and a Naïve Bayes-based score-fuser.

3D Facial Matching by Spiral Convolutional Metric Learning and a Biometric Fusion-Net of Demographic Properties


This paper presents a novel Reducing-Over-Time (ROT) binary tree structure for event-based vision data and sub-types of the tree structure. A framework is presented using ROT, that takes advantage of the self-balancing and self-pruning nature of the tree structure to extract spatial-temporal information. The ROT framework is paired with an established motion classification technique and performance is evaluated against other state-of-the-art techniques using four datasets. Additionally, the ROT framework as a processing platform is compared with other event-based vision processing platforms in terms of memory usage and is found to be one of the most memory efficient platforms available.

Reducing-Over-Time Tree

Gaussian Process Models (GPMs) are Bayesian machine learning models that have been widely applied in the domain of pattern recognition due to their ability to infer from unreliable, noisy, or highly idiosyncratic data. Retrieving a complex GPM describing the data’s inherent statistical patterns, such as trends, seasonalities, and periodicities, is a key requirement for various pattern recognition tasks. In this paper, we propose a novel approach for efficient large-scale GPM retrieval: the Concatenated Composite Covariance Search (3CS) algorithm. By making use of multiple local kernel searches on dynamically partitioned data, the 3CS algorithm is able to overcome the performance limitations of state-of-the-art GPM retrieval algorithms and to efficiently retrieve GPMs for largescale data up to three orders of magnitude as fast as state-ofthe- art algorithms.

3CS Algorithm for Efficient Gaussian Process Model Retrieval

Deep learning technology has enabled successful modeling of complex facial features when high quality images are available. Nonetheless, accurate modeling and recognition of human faces in real world scenarios 'on the wild' or under adverse conditions remains an open problem. When unconstrained faces are mapped into deep features, variations such as illumination, pose, occlusion, etc., can create inconsistencies in the resultant feature space. Hence, deriving conclusions based on direct associations could lead to degraded performance. This rises the requirement for a basic feature space analysis prior to face recognition. This paper devises a joint clustering and classification scheme which learns deep face associations in an easy-to-hard way. Our method is based on hierarchical clustering where the early iterations tend to preserve high reliability. The rationale of our method is that a reliable clustering result can provide insights on the distribution of the feature space, that can guide the classification that follows. Experimental evaluations on three tasks, face verification, face identification and rank-order search, demonstrates better or competitive performance compared to the state-of-the-art, on all three experiments.

ClusterFace: Joint Clustering and Classification for Set-Based Face Recognition

Autonomous driving of a personal mobility vehicle such as a wheelchair in a walking space is crucial as a means of transportation for the elderly and the physically handicapped. To realize this, accurate pedestrian detection is indispensable. As existing 3D object detection methods are trained with a roadway dataset, they are widely used for object detection in roadways. These methods have two major drawbacks as regards the detection of objects in walking spaces. The first is that they largely depend on the different LIDAR models. To eliminate this issue, we propose a 3D object detection method, CosPointPillars, that does not take the reflection intensities of the LIDAR point cloud, which causes a sensor model dependency, as input. The second drawback is that networks trained with a roadway dataset cannot sufficiently detect pedestrians (who are major traffic participants in walking spaces) located within a short distance; this is because the roadway dataset hardly includes nearby pedestrians. To solve this issue, we generated a new walking space dataset called SimDataset, which includes nearby pedestrians as a training dataset in the simulations. An experiment on a real walking space showed that SimDataset is suitable for use in pedestrian detection.

Sensor-independent Pedestrian Detection for Personal Mobility Vehicles in Walking Space Using Dataset Generated by Simulation

Unsupervised domain adaptation has achieved significant results by leveraging knowledge from a source domain to learn a related but unlabeled target domain. Previous methods are insufficient to model domain discrepancy and class discrepancy, which may lead to misalignment and poor adaptation performance. To address this problem, in this paper, we propose a unified framework, called distance-aware domain adaptation, which is fully aware of both cross-domain distance and class-discriminative distance. In addition, second-order statistics distance and manifold alignment are also exploited to extract more information from data. In this manner, the generalization error of the target domain in classification problems can be reduced substantially. To validate the proposed method, we conducted experiments on five public datasets and an ablation study. The results demonstrate the good performance of our proposed method.

A Unified Framework for Distance-Aware Domain Adaptation


In this paper, a new recursive structure based
on the convolution model of discrete cosine transform (DCT)
for designing of a finite impulse response (FIR) digital filter
is proposed. In our derivation, we start with the convolution
model of DCT-II to use its Z-transform for the proposed filter
structure perspective. Moreover, using the same algorithm, a
filter base implementation of the inverse DCT (IDCT) for image
reconstruction is developed. The computational time experiments
of the proposed DCT/IDCT filter(s) demonstrate that the proposed
filters achieve faster elapsed CPU time compared to the
others. The image filtering and reconstruction performance of
the proposed approach on ultrasound images are presented to
validate the theoretical framework.

DCT/IDCT Filter Design for Ultrasound Image Filtering

We study the problem of unsupervised domain adaptation that aims at obtaining a prediction model for the target domain using labeled data from the source domain and unlabeled data from the target domain. There exists an array of recent research based on the idea of extracting features that are not only invariant for both domains but also provide high discriminative power for the target domain. In this paper, we propose an idea of improving the discriminativeness: Adding an extra artificial class and training the model on the given data together with the GAN-generated samples of the new class. The trained model based on the new class samples is capable of extracting the features that are more discriminative by repositioning data of current classes in the target domain and therefore increasing the distances among the target clusters in the feature space. Our idea is highly generic so that it is compatible with many existing methods such as DANN, VADA, and DIRT-T. We conduct various experiments for the standard data commonly used for the evaluation of unsupervised domain adaptations and demonstrate that our algorithm achieves the SOTA performance for many scenarios.

Enlarging Discriminative Power by Adding an Extra Class in Unsupervised Domain Adaptation

Estimated Time of Arrival (ETA) is one of the most fundamental problems in Intelligent Transportation System. It is considered as predicting the travel time from the origin to the destination along a given path. The route consists of a sequence of links. One of the key techniques is to use embedding vectors to represent links. However, the embedding suffers from the data sparsity problem that many links are traversed by too few floating cars even in large ride-hailing platforms. To address the problem, we propose the Road Network Metric Learning framework for ETA (RNML-ETA). We transfer the knowledge of hot links to the cold links by metric learning. In this presentation, we will introduce the key points of RNML-ETA which is the first deep learning method that effectively addresses the road network data sparsity problem. On the large scale real-world datasets, our method outperforms the state-of-the-art model and the promotion concentrates on the cold links with few data.

Road Network Metric Learning for Estimated Time of Arrival


Recent works have made great progress in semantic segmentation by exploiting contextual information in a local or global manner with dilated convolutions, pyramid pooling or self-attention mechanism. However, few works have focused on harvesting boundary information to improve the segmentation performance. In order to enhance the feature similarity within the object and keep discrimination from other objects, we propose a boundary-aware graph convolution (BGC) module to propagate features within the object. The graph reasoning is performed among pixels of the same object apart from the boundary pixels. Based on the proposed BGC module, we further introduce the Boundary-aware Graph Convolution Network(BGCNet), which consists of two main components including a basic segmentation network and the BGC module, forming a coarse-to-fine paradigm. Specifically, the BGC module takes the coarse segmentation feature map as node features and boundary prediction to guide graph construction. After graph convolution, the reasoned feature and the input feature are fused together to get the refined feature, producing the refined segmentation result. We conduct extensive experiments on three popular semantic segmentation benchmarks including Cityscapes, PASCAL VOC 2012 and COCO Stuff, and achieve state-of-the-art performance on all three benchmarks.


Downloads

Next from ICPR 2020

3D Facial Matching by Spiral Convolutional Metric Learning and a Biometric Fusion-Net of Demographic Properties

Similar lecture

A Two-Stream Recurrent Network for Skeleton-based Human Interaction Recognition