Italy


Scene text detection has been studied extensively. Existing methods detect either words or text lines and use either word-level or line-level annotated data for training. In this work, we propose a dual-task network that can perform word-level and line-level text detection simultaneously and use training data of both levels of annotation to boost the performance. The dual-task network has two detection heads for word-level and line-level text detection, respectively. Then we propose a mutual guidance scheme for the joint training of the two tasks with two modules: line filtering module utilizes the output feature map of the text line detector to filter out the non-text regions for the word detector, and word enhancing module provides prior positions of words for the text line detector depending on the output feature map of the word detector. Experimental results of word-level and linelevel text detection demonstrate the effectiveness of the proposed dual-task network and mutual guidance scheme, and the results of our method are competitive with state-of-the-art methods.

ICPR 2020

Mutually Guided Dual-Task Network for Scene Text Detection

mutual guidance

dual-task network

scene text detection

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. 


technical paper


For the last decade, single image defogging has been a subject of interest in image processing. In the visible spectrum, fog and haze decrease the visibility of distant objects. Thus, the objective of the visibility restoration is to remove as much as possible the effect of the fog within the image. Infrared sensors are more and more used in automotive and aviation industries but the effect of fog and haze is not restricted to the visible spectrum and also applies in the infrared band. After recalling the effects of fog in the common sub-bands of the infrared spectrum, we tested if the approach used for single image defogging in the visible spectrum might also work for infrared. This led us to propose a new approach of single image defogging for Long-Wavelength Infra-Red (LWIR) or Thermal Infra-Red. Several experiments are presented showing that the proposed algorithm offers interesting results not only for fog and haze but for bad weather conditions in general, during day and night.

Visibility Restoration in Infra-Red Images

We propose a novel foreground-focused domain adaptation (FFDA) framework
which mines the loss of the domain discriminators to concentrate on the backpropagation of foreground loss.

Foreground-focused domain adaptation for object detection


Understanding images by predicting pixel-level semantic classes is a fundamental task in computer vision and is one of the most important techniques for autonomous driving. Recent approaches based on deep convolutional neural networks have dramatically improved the speed and accuracy of semantic segmentation on paved road datasets, however, dirt roads have yet to be systematically studied. Dirt roads do not contain clear boundaries between drivable and non-drivable regions; and thus, this difficulty must be overcome for the realization of fully autonomous vehicles. The key idea of our approach is to apply lightweight non-local blocks to reinforce stage-wise long-range dependencies in encoder-decoder style backbone networks. Experiments on 4,687 images of a dirt road dataset show that our transitional asymmetric non-local neural networks present a higher accuracy with lower computational costs compared to state-of-the-art models.

Transitional Asymmetric Non-local Neural Network for Real-Time Dirt Road Segmentation

This paper presents a novel active learning technique aimed
at the selection of sequences for manual annotation from a
database of unlabelled sequences. Supervised machine
learning algorithms can employ these sequences to build
better models than those based on using random sequences
for training. The main contribution of the proposed method
is the use of sequential pattern information contained in
the database to select representative and diverse sequences
for annotation. These two characteristics ensure the proper
coverage of the instance space of sequences and, at the
same time, avoids over-fitting the trained model. The
approach, called SPIAL (Sequential Pattern Information for
Active Learning), uses sequential pattern mining algorithms
to extract frequently occurring sub-sequences from the
database and evaluates how representative and diverse each
sequence is, based on this information. The output is a
list of sequences for annotation sorted by
representativeness and diversity. The algorithm is modular
and, unlike most current techniques, it is independent from
the sequential machine learning technique that builds the
model as SPIAL does not need to know the set of the
features taken into account by the training algorithm.
Experiments done on well-known benchmarks involving
sequential data show that the models trained using SPIAL
increase their convergence speed while reducing manual
effort by selecting small sets of very informative
sequences for annotation. In addition, the computation cost
using SPIAL is much lower than for the state-of-the-art
algorithms evaluated.

Leveraging Sequential PAttern Information for Active Learning from Sequential Data


Obituaries represent a prominent way of expressing the human universal of grief. According to philosophers, obituaries are a ritualized way of evaluating both individuals who have passed away and the communities that helped to shape them. The basic idea is that you can tell what it takes to count as a good person of a particular type in a particular community by seeing how persons of that type are described and celebrated in their obituaries. Obituaries of those killed in conflict, in particular, are rich repositories of communal values, as they reflect the values and virtues that are admired and respected in individuals who are considered to be heroes in their communities. In this paper, we use natural language processing techniques to map the patterns of values and virtues attributed to Australian military personnel who were killed in action during World War I and World War II. Doing so reveals several clusters of values and virtues that tend to be attributed together. In addition, we use named entity recognition and geotagging to track the movements of these soldiers to various theatres of the wars, including North Africa, Europe, and the Pacific.

To Honor our Heroes: Analysis of the Obituaries of Australians Killed in Action in WWI and WWII

3D face reconstruction from a single 2D facial image is a challenging and concerned problem. Recent methods based on CNN typically aim to learn parameters of 3D Morphable Model (3DMM) from 2D images to render face alignment and 3D face reconstruction. Most algorithms are designed for faces with small, medium yaw angles, which is extremely challenging to align faces in large poses. At the same time, they are not efficient usually. The main challenge is that it takes time to determine the parameters accurately. In order to address this challenge with the goal of improving performance, this paper proposes a novel and efficient end-to-end framework. We design an efficient and lightweight network model combined with Depthwise Separable Convolution and Muti-scale Representation, Lightweight Attention Mechanism, named Mobile-FRNet. Simultaneously, different loss functions are used to constrain and optimize 3DMM parameters and 3D vertices during training to improve the performance of the network. Meanwhile, extensive experiments on the challenging datasets show that our method significantly improves the accuracy of face alignment and 3D face reconstruction. Model parameters and complexity of our method are also improved greatly.

Joint Face Alignment and 3D Face Reconstruction with Efficient Convolution Neural Networks

Tensor completion and robust principal component analysis have been widely used in machine learning while the key problem relies on the minimization of a tensor rank that is very challenging. A common way to tackle this difficulty is to approximate the tensor rank with the $\ell_1-$norm of singular values based on its Tensor Singular Value Decomposition (T-SVD). Besides, the sparsity of a tensor is also measured by its $\ell_1-$norm. However, the $\ell_1$ penalty is essentially biased and thus the result will deviate. In order to sidestep the bias, we propose a novel non-convex tensor rank surrogate function and a novel non-convex sparsity measure. In this new setting by using the concavity instead of the convexity, a majorization minimization algorithm is further designed for tensor completion and robust principal component analysis. Furthermore, we analyze its theoretical properties. Finally, the experiments on natural and hyperspectral images demonstrate the efficacy and efficiency of our proposed method.

T-SVD Based Non-convex Tensor Completion and Robust Principal Component Analysis


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.

On the Information of Feature Maps and Pruning of Deep Neural Networks

Person re-identification (reid) aims at matching pedestrians observed from non-overlapping camera views. It has important applications in surveillance video analysis such as human retrieval, human tracking and activity analysis. Although a large number of effective feature learning and distance metric optimizing approaches have been proposed, it still suffers from pedestrian appearance variations caused by pose changing. Most of the previous methods address this problem by learning a pose-invariant descriptor subspace. In this paper, we propose a pose variation adaptation method for person reid. It can reduce the probability of deep learning network over-fitting. Specifically, we introduce a pose transfer generative adversarial network with a similarity measurement module. With the learned pose transfer model, training images can be transferred to any given poses, and with the original images, forming an augmented training dataset. It increases data diversity against over-fitting. In contrast to previous GAN-based methods, we consider the influence of pose variations on similarity measures to generate shaper and more realistic samples for person reid. Besides, we optimize hard example mining to introduce a novel manner of samples used with the learned pose transfer model. It focuses on the inferior samples which are caused by pose variations to increase the number of effective hard examples for learning discriminative features and improving the generalization ability. We extensively conduct comparative evaluations to demonstrate the advantages and superiorities of the proposed method over the state-of-the-art person reid approaches on Market-1501 and DukeMTMC-reID.

Pose Variation Adaptation for Person Re-identification

Eye movements are intricate and dynamic events that contain a wealth of information about the subject and the stimuli. We propose an abstract representation of eye movements that preserve the important nuances in gaze behavior while being stimuli-agnostic. We consider eye movements as raw position and velocity signals and train separate deep temporal convolutional autoencoders. The autoencoders learn micro-scale and macro-scale representations that correspond to the fast and slow features of eye movements. We evaluate the joint representations with a linear classifier fitted on various classification tasks. Our work accurately discriminates between gender and age groups and outperforms previous works on biometrics and stimuli classification. Further experiments highlight the validity and generalizability of this method, bringing eye-tracking research closer to real-world applications.

Premium content

Mutually Guided Dual-Task Network for Scene Text Detection

Downloads

Next from ICPR 2020

Visibility Restoration in Infra-Red Images

Similar lecture

Mutually Guided Dual-Task Network for Scene Text Detection