Italy

Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human pose estimation algorithms typically produce much less accurate feature points compared to classical patch-based methods. Another problem is that ambient human motion might not be optimal for calibration. We build upon prior works and introduce several novel ideas to improve the accuracy of human-pose-based extrinsic calibration. Our first contribution is a robust reprojection loss based on a better understanding of the sources of pose estimation error. Our second contribution is a 3D human pose likelihood model learned from motion capture data. We demonstrate significant improvements in calibration accuracy by evaluating our method on four publicly available datasets.

ICPR 2020

Better Prior Knowledge Improves Human-Pose-Based Extrinsic Camera Calibration

cameras

pose estimation

calibration

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


Ground Terrain Recognition is a difficult task as the context information varies significantly over the regions of a ground terrain image. In this paper, we propose a novel approach towards ground-terrain recognition via modeling the Extent-of-Texture information to establish a balance between the orderless texture component and ordered-spatial information locally. At first, the proposed method uses a CNN backbone feature extractor network to capture meaningful information of a ground terrain image, and model the extent of texture and shape information locally. Then, the order-less texture information and ordered shape information are encoded in a patch-wise manner, which is utilized by intra-domain message passing module to make every patch aware of each other for rich feature learning. Next, the Extent-of-Texture (EoT) Guided Inter-domain Message Passing module combines the extent of texture and shape information with the encoded texture and shape information in a patchwise fashion for sharing knowledge to balance out the order-less texture information with ordered shape information. Further, Bilinear model generates a pairwise correlation between the order-less texture information and ordered shape information. Finally, the ground-terrain image classification is performed by a fully connected layer. The experimental results indicate superior performance of the proposed model over existing state-of-the-art techniques on publicly available datasets like DTD, MINC and GTOS-mobile.

Modeling Extent-of-Texture Information for Ground Terrain Recognition

Ultrasound-Guided Regional Anesthesia (UGRA) becomes a standard procedure in surgical operations and contributes to pain management. It offers the advantages of targeted nerve detection and provides the visualization of regions of interest such as anatomical structures. However, nerve detection is one of the most challenging tasks that anesthetists can encounter in the UGRA procedure. A computer-aided system that can detect automatically the nerve region would facilitate the anesthetist’s daily routine and allow them to concentrate more on the anesthetic delivery. In this paper, we propose a new method based on merging deep learning models from different data to detect the median nerve. The merged architecture consists of two branches, one being one-dimensional (1D) convolutional neural networks (CNN) branch and another 2D CNN branch. The merged architecture aims to learn the high-level features from 1D handcrafted noise-robust features and 2D ultrasound images. The obtained results show the validity and high accuracy of the proposed approach and its robustness.

Merged 1D-2D Deep Convolutional Neural Networks for Nerve Detection in Ultrasound Images

I will introduce an image-level solution for action detection which especially is suitable for professional action localization. With a small amount of videos, our framework can achieve high temporal localization precision. In addition, our model can naturally avoid detections of incomplete action.

Precise Temporal Localization for Complete Actions with Quantified Temporal Structure

Neural Architecture Search (NAS) has gained attraction due
to superior classification performance. Differential
Architecture Search (DARTS) is a computationally light
method. To limit computational resources DARTS makes
numerous approximations. These approximations result in
inferior performance. We propose to fine-tune DARTS using
fixed operations as they are independent of these
approximations. Our method offers a good trade-off between
the number of parameters and classification accuracy. Our
approach improves the top-1 accuracy on Fashion-MNIST,
CompCars and MIO-TCD datasets by 0.56%, 0.50%, and 0.39%,
respectively compared to the state-of-the-art approaches.
Our approach performs better than DARTS, improving the
accuracy by 0.28%, 1.64%, 0.34%, 4.5%, and 3.27% compared
to DARTS, on CIFAR-10, CIFAR-100, Fashion-MNIST, CompCars,
and MIO-TCD datasets, respectively.

Fine-Tuning DARTS for Image Classification

Deep neural networks forget previously learnt tasks when they are faced with learning new tasks. This is called catastrophic forgetting. Rehearsing the neural network with the training data of the previous task can protect the network from catastrophic forgetting. Since rehearsing requires the storage of entire previous data, Pseudo rehearsal was proposed, where samples belonging to the previous data are generated synthetically for rehearsal. In an image classification setting, while current techniques try to generate synthetic data that is photo-realistic, we demonstrated that Neural networks can be rehearsed on data that is not photo-realistic and still achieve good retention of the previous task. We also demonstrated that forgoing the constraint of having photo realism in the generated data can result in a significant reduction in the consumption of computational and memory resources for pseudo rehearsal.

Pseudo rehearsal using non photo-realistic images


Environmental protection is one of the main challenges facing society nowadays. Even with constantly growing awareness, not all of the sorting can be done by people themselves - the differences between materials are not visible to the human eye. For that reason, we present the use of a hyperspectral camera as a capture device, which allows us to obtain the full spectrum of the material. In this work, we propose a method for efficient recognition of the substance of an item. We conducted several experiments and analysis of the spectra of different materials in different conditions on a special measuring stand. That enabled the identification of the best features, which can later be used during classification, which was confirmed during the extensive testing procedure.

Hyperspectral Imaging for Analysis and Classification of Plastic Waste


We present a new encoding method LODEC and a new model LODENet for Chinese/Japanese OCR.
The results achieve SOTA results on CASIA, SCUT-EPT, and a Japanese dataset.

LODENet: A Holistic Approach to Offline Handwritten Chinese and Japanese Text Line Recognition

Several typical types of retinopathy are major causes
of blindness. However, early detection of retinopathy is quite not easy since few symptoms are observable in the early stage, attributing to the development of non-mydriatic retinal cameras, these cameras produce high-resolution retinal fundus images that provide the possibility of Computer-Aided-Diagnosis (CAD) via deep learning to assist diagnosing retinopathy. Deep learning algorithms usually rely on a large number of labeled images that are expensive and time-consuming to obtain in the medical imaging area. Moreover, the random distribution of various lesions that often vary greatly in size also brings significant challenges to learn discriminative information from high-resolution fundus images. In this paper, we present generative adversarial networks simultaneously equipped with a ”good” generator and a ”bad” generator (GBGANs) to make up for the incomplete data distribution given limited fundus images. To improve the generative feasibility of the generator, we introduce a pre-trained feature extractor to acquire condensed features for each fundus image in advance. Experimental results on integrated three public iChallenge datasets show that the proposed GBGANs could fully utilize the available fundus images to identify retinopathy with little label cost.

Semi-Supervised Generative Adversarial Networks with a Pair of Complementary Generators for Retinopathy Screening


The recent progress of deep convolutional neural networks has enabled great success in single image super-resolution (SISR) and many other vision tasks. Their performances are also being increased by deepening the networks and developing more sophisticated network structures. However, finding an optimal structure for the given problem is a difficult task, even for human experts. For this reason, neural architecture search (NAS) methods have been introduced, which automate the procedure of constructing the structures. In this paper, we expand the NAS to the super-resolution domain and find a lightweight densely connected network named DeCoNASNet. We use a hierarchical search strategy to find the best connection with local and global features. In this process, we define a complexity-based penalty for solving image super-resolution, which can be considered a multi-objective problem. Experiments show that our DeCoNASNet outperforms the state-of-the-art lightweight super-resolution networks designed by handcraft methods and existing NAS-based design.

1325 - Neural Architecture Search for Image Super-Resolution Using Densely Connected Search Space: DeCoNAS

The “2019 Automatic Speaker Verification Spoofing And Countermeasures Challenge” (ASVspoof) competition aimed to facilitate the design of highly accurate voice spoofing attack detection systems. the competition did not emphasize model
complexity and latency requirements; such constraints are strict and integral in real-world deployment. Hence, most of the top performing solutions from the competition all used an ensemble approach, and combined multiple complex deep learning models to maximize detection accuracy – this kind of approach would sit uneasily with real-world deployment constraints. To design a lightweight system, we combined the notions of skip connection (from ResNet) and max feature map (from Light CNN), and evaluated the accuracy of the system using the ASVspoof 2019
dataset. With an optimized constant Q transform (CQT) feature, our single model achieved a replay attack detection equal error rate (EER) of 0.37% on the evaluation set, surpassing the top ensemble system from the competition that achieved an EER of 0.39%

Downloads

Next from ICPR 2020

Modeling Extent-of-Texture Information for Ground Terrain Recognition

Similar lecture

StrongPose: Bottom-up and Strong Keypoint Heat Map Based Pose Estimation