Italy

Human capability to anticipate near future from visual observations and non-verbal cues is essential for developing intelligent systems that need to interact with people. Several research areas, such as human-robot interaction (HRI), assisted living or autonomous driving need to foresee future events to avoid crashes or help people. Egocentric scenarios are classic examples where action anticipation is applied due to their numerous applications. Such challenging task demands to capture and model domain’s hidden structure to reduce prediction uncertainty. Since multiple actions may equally occur in the future, we treat action anticipation as a multi-label problem with missing labels extending the concept of label smoothing. This idea resembles the knowledge distillation process since useful information is injected into the model during training. We implement a multi-modal framework based on long short-term memory (LSTM) networks to summarize past observations and make predictions at different time steps. We perform extensive experiments on EPIC-Kitchens and EGTEA Gaze+ datasets including more than 2500 and 100 action classes, respectively. The experiments show that label smoothing systematically improves performance of state-of-the-art models for action anticipation.

ICPR 2020

Knowledge Distillation for Action Anticipation via Label Smoothing

action anticipation

label smoothing

knowledge distillation

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


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.

Conditional Multi-task learning for Plant Disease Identification


Scene text recognition has been a hot topic in computer vision. Recent methods adopt the attention mechanism for sequence prediction which achieve convincing results. However, we argue that the existing attention mechanism faces the problem of attention diffusion, in which the model may not focus on a certain character area. In this paper, we propose Gaussian Constrained Attention Network to deal with this problem. It is a 2D attention-based method integrated with a novel Gaussian Constrained Refinement Module, which predicts an additional Gaussian mask to refine the attention weights. Different from adopting an additional supervision on the attention weights simply, our proposed method introduces an explicit refinement. In this way, the attention weights will be more concentrated and the attention-based recognition network achieves better performance. The proposed Gaussian Constrained Refinement Module is flexible and can be applied to existing attention-based methods directly. The experiments on several benchmark datasets demonstrate the effectiveness of our proposed method.

Gaussian Constrained Attention Network for Scene Text Recognition

In this paper, we make the assumption that an expression with a higher accuracy is easier to be recognized, and those expressions easier to recognize will hinder the recognition of uneasy expressions. Then, we propose a novel algorithm for unconstrained FER based on cascade decision and Gabor filters. Easier expressions are recognized before the difficult expressions. This simple method trains up to five models to cascadedly recognize a given facial image expression. Gabor filters are included in every model to enhance robustness on illumination variations and face poses. Extensive experiment results on several datasets validate the effectiveness of the proposed method. We obtain accuracy of 77.6% on FER2013 with the final models, outperforming the latest state-of-the-arts.

Unconstrained Facial Expression Recogniton Based on Cascade Decision and Gabor Filters

Deep Convolutional Neural Networks (DCNN), despite imitating the human visual system, present no such decision credibility as human observers. This phenomenon, therefore, leads to the limitations of DCNN's applications in the security and trusted computing, such as self-driving cars and medical diagnosis. Focusing on this issue, our work aims to explore the way DCNN makes decisions. In this paper, the major contributions we made are: firstly, provide the hypothesis, “point-wise activation” of convolution function, according to the analysis of DCNN’s architectures and training process; secondly, point out the effect of “point-wise activation” on DCNN’s uninterpretable classification and pool robustness, and then suggest, in particular, the contradiction between the traditional and DCNN’s convolution kernel functions; finally, distinguish decision-making interpretability from semantic interpretability, and indicate that DCNN’s decision-making mechanism need to evolve towards the direction of semantics in the future. Besides, the “point-wise activation” hypothesis and conclusions proposed in our paper are supported by extensive experimental results.

How Does DCNN Make Decisions?

It is difficult to achieve high performance without sufficient training data for deep convolutional neural networks (DCNNs) to learn. Data augmentation plays an important role in improving robustness and preventing overfitting in machine learning for many applications such as image classification. In this paper, a novel method for data augmentation is proposed to solve the problem of machine learning with small training datasets. The proposed method can synthesise similar images with rich diversity from only a single original training sample to increase the number of training data by using generative adversarial networks (GANs). It is expected that the synthesised images possess class-informative features, which may be in the validation or testing data but not in the training data due to that the training dataset is small, and thus they can be effective as augmented training data to improve the classification accuracy of DCNNs. The experimental results have demonstrated that the proposed method with a novel GAN framework for image training data augmentation can significantly enhance the classification performance of DCNNs for applications where original training data is limited.

Augmentation of Small Training Data Using GANs for Enhancing the Performance of Image Classification


Deep learning approaches have achieved impressive results in breast ultrasound (BUS) image segmentation. However, these methods did not solve uncertainty and noise in BUS images well. Meanwhile, they did not involve the context information of BUS images, either. To address this issue, we present a novel deep learning structure for BUS image semantic segmentation by analyzing the uncertainty using a pyramid fuzzy block and generating a novel feature based on connectedness. There are three major contributions in this paper: (1) the structure of pyramid fuzzy block; (2) a novel membership function based on multi-convolution layers; and (3) a novel context feature based on connectedness. The proposed methods are applied to two datasets: a BUS image benchmark with two categories (background and tumor) and a five-category BUS image dataset with fat layer, mammary layer, muscle layer, background, and tumor. The proposed method achieves the best results on both datasets compared with eight state-of-the-art deep learning-based approaches.

Semantic Segmentation of Breast Ultrasound Image with Pyramid Fuzzy Uncertainty Reduction and Direction Connectedness Feature

We focus on 1 : N identification tasks with binary features. 
Most multiclass classification methods, including identification and verification methods, use a shared metric space in which distances between samples are measured regardless of their identities. This is because dedicated metric spaces learned for each identity in the training set are of little use for the test set. In 1 : N identification tasks, however, gallery samples contain rich information about the test domain. Given a sample and its neighbors in the gallery set, we propose a method for calculating a discriminative feature selection mask that is used as a sample-dependent distance metric. Experiments on several reidentification datasets show that the proposed method enhances the performance of state-of-the-art feature extractors.

Sample-Dependent Distance for 1 : N Identification via Discriminative Feature Selection

In this presentation, we focus on evaluating the initialization of the Nelder-Mead method for hyperparameter optimization in deep learning. The Nelder-Mead (NM) method is known for superior performance for hyperparameter optimization in deep learning. An initial value, one of the initial NM method’s values, is usually determined randomly while the search performance strongly depends on the initial value. Therefore, we clarified its influence on the search performance to determine the proper initial value.
The result contributes to the determination of the initial value in the Nelder-Mead method.

Evaluating Initialization of Nelder-Mead Method for Hyperparameter Optimization in Deep Learning

With the development of technology, biometrics are used for identification. As a biological behavior feature, lip motion has become a new kind of biometric in speaker recognition. In the literature, lots of works used 2D lip images to recognize speaker in a text-dependent context. However, 2D lip easily suffers from face orientations. To this end, in this work, we present a novel end-to-end 3D lip motion Network (3LMNet) by utilizing the sentence-level 3D lip motion (S3DLM) to recognize speakers in both the text-independent and text-dependent contexts.

Three-Dimensional Lip Motion Network for Text-Independent Speaker Recognition


Attention module has achieved great success in numerous vision tasks. However, existing visual attention modules generally consider the features of a single-scale, and cannot make full use of their multi-scale contextual information. Meanwhile, the multi-scale spatial feature representation has demonstrated its outstanding performance in a wide range of applications. However, the multi-scale features are always represented in a layer-wise manner, i.e. it is impossible to know their contextual information at a granular level. Focusing on the above issue, a context-aware residual module for image classification is proposed in this paper. It consists of a novel multi-scale channel attention module MSCAM to learn refined channel weights by considering the visual features of its own scale and its surrounding fields, and a multi-scale spatial aware module MSSAM to further capture more spatial information. Either or both of the two modules can be plugged into any CNN-based backbone image classification architecture with a short residual connection to obtain the context-aware enhanced features. The experiments on public image recognition datasets including CIFAR10, CIFAR100，Tiny-ImageNet and ImageNet consistently demonstrate that our proposed modules significantly outperforms a wide-used state-of-the-art methods, e.g., ResNet and the lightweight networks of MobileNet and SqueezeeNet.

Premium content

Knowledge Distillation for Action Anticipation via Label Smoothing

Downloads

Next from ICPR 2020

Conditional Multi-task learning for Plant Disease Identification

Similar lecture

Channel Planting for Deep Neural Networks Using Knowledge Distillation