Italy


To understand human behavior we must not just recognize individual actions but model possibly complex group activity and interactions. Hierarchical models obtain the best results in group activity recognition but require fine grained individual action annotations at the actor level. We show that, considering only skeletal data, we can train a state-of-the art end-to-end system using only group activity labels at the sequence level.

ICPR 2020

Learning Group Activities from Skeletons without Individual Action Labels

group activity recognition

action recognition

semi supervised learning

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

This paper proposes the first object counting method for omnidirectional images. Because conventional object counting methods cannot handle the distortion of omnidirectional images, we propose to process them using stereographic projection,
which enables conventional methods to obtain a good approximation of the density function. However, the images obtained by stereographic projection are still distorted. Hence, to manage this distortion, we propose two methods. One is a new data augmentation method designed for the stereographic projection of omnidirectional images. The other is a distortion-adaptive Gaussian kernel that generates a density map ground truth while taking into account the distortion of stereographic projection. Using the counting of grape bunches as a case study, we
constructed an original grape-bunch image dataset consisting of omnidirectional images and conducted experiments to evaluate the proposed method. The results show that the proposed method performs better than a direct application of the conventional method, improving mean absolute error by 14.7% and mean squared error by 10.5%.

Distortion-Adaptive Grape Bunch Counting for Omnidirectional Images

Given a face image of a recovered child at probe-age, we
search a gallery of missing children with known identities
and gallery-ages at which they were either lost or stolen
in an attempt to unite the recovered child with his family.
We propose a feature aging module that can age-progress
deep face features output by a face matcher to improve the
recognition accuracy of age-separated child face images. In
addition, the feature aging module guides age-progression
in the image space such that synthesized aged gallery faces
can be utilized to further enhance cross-age face matching
accuracy of any commodity face matcher. For time lapses
larger than 10 years (the missing child is recovered after
10 or more years), the proposed age-progression module
improves the rank-1 open-set identification accuracy of
CosFace from 22.91% to 25.04% on a child celebrity dataset,
namely ITWCC. The proposed method also outperforms
state-of-the-art approaches with a rank-1 identification
rate of 95.91%, compared to 94.91%, on a public aging
dataset, FG-NET, and 99.58%, compared to 99.50%, on
CACD-VS. These results suggest that aging face features
enhances the ability to identify young children who are
possible victims of child trafficking or abduction.

Identifying Missing Children: Face Age Progression via Deep Feature Aging

Recent advances in computer vision and machine learning, most notably deep convolutional neural networks (CNNs), are exploited to identify and localize various plant species in salt marsh images. Three different approaches are explored that provide estimations of abundance and spatial distribution at varying levels of granularity defined by spatial resolution. In the coarsest-grained approach, CNNs are tasked with identifying which of six plant species are present/absent in large patches within the salt marsh images. CNNs with diverse topological properties and attention mechanisms are shown capable of providing accurate estimations with > 90% precision and recall for the more abundant plant species and reduced performance for less common plant species. Estimation of percent cover of each plant species is performed at a finer spatial resolution, where smaller image patches are extracted and the CNNs tasked with identifying the plant species or substrate at the center of the image patch. For the percent cover estimation task, the CNNs are observed to exhibit a performance profile similar to that for the presence/absence estimation task, but with a 5%-10% reduction in precision and recall. Finally, fine-grained estimation of the spatial distribution of the various plant species is performed via semantic segmentation. The DeepLab-V3 semantic segmentation architecture is observed to provide very accurate estimations for abundant plant species, but with significant performance degradation for less abundant plant species; in extreme cases, rare plant classes are seen to be ignored entirely. Overall, a clear trade-off is observed between the CNN estimation quality and the spatial resolution of the underlying estimation thereby offering guidance for ecological applications of CNN-based approaches to automated plant identification and localization in salt marsh images.

Estimation of Abundance and Distribution of Salt Marsh Plants from Images Using Deep Learning


We present a new framework for learning-free 3D human pose estimation from Inertial Measurement Unit (IMU) data. The proposed method does not rely on a full motion sequence to calculate a pose for any particular time point and thus can operate in real-time. A cost function based only on joint rotations is used, removing the need for frequent transformations between rotations and 3D Cartesian coordinates. A Jacobian that preserves skeleton structure is derived using Adjoint methods from Variational Data Assimilation. To facilitate further research in IMU-based Motion Capture, we provide a dataset that combines RGB and depth images from an Intel RealSense camera, marker-based motion capture from an Optitrack system and Xsens IMU data. We have evaluated our method on both our dataset and the Total Capture dataset, showing an average error across 24 joints of 0.45 and 0.48 radians respectively.

Rotational Adjoint Methods for Learning-Free 3D Human Pose Estimation from IMU Data

Recent advances in unmanned aerial vehicles and camera technology have proven useful for the detection of smoke that emerges above the trees during a forest fire. Automatic detection of smoke in videos is of great interest to Fire department. To date, in most parts of the world, the fire is not detected in its early stage and generally it turns catastrophic. The proposed work introduces a novel technique that integrates spatial and temporal features in a deep learning framework using semi supervised spatio-temporal video object segmentation and dense optical flow. However, detecting this smoke in the presence of haze and without the labeled data is difficult. Considering the visibility of haze in the sky, a dark channel pre-processing method is used that reduces the amount of haze in video frames and consequently improves the detection results. Online training is performed on a video at the time of testing that reduces the need for ground-truth data. Tests using the publicly available video datasets show that the proposed algorithms outperform previous work and they are robust across different wildfire-threatened locations.

Early Wildfire Smoke Detection in Videos

Over 40% of the world’s population is bilingual. Existing speaker identification/verification systems, however, as- sume the same language type for both enrollment and recognition stages. In this work, we investigate the feasibility of employing multilingual speech for biometric application. We establish a dataset containing audio recorded in English, Mandarin and Tai- wanese. Three acoustic features, namely, i-vector, d-vector and x- vector have been evaluated for both speaker verification (SV) and identification (SI) tasks. Preliminary experimental results indicate that x-vector achieves the best overall performance. Additionally, model trained with hybrid data demonstrates highest accuracy associated with the cost of data collection efforts.
In SI tasks, we obtained over 91% cross-lingual accuracy all models using 3-second audio. In SV tasks, the EER among cross- lingual test is at most 6.52%, which is observed on the model trained by English corpus. The outcome suggests the feasibility of adopting cross-lingual speech in building text-independent speaker recognition systems.

Toward Text-independent Cross-lingual Speaker Recognition Using English-Mandarin-Taiwanese Dataset


Recent advances in camera equipped drone applications and their widespread use increased the demand on vision based object detection algorithms for aerial images. Object detection process is inherently a challenging task as a generic computer vision problem, however, since the use of object detection algorithms on UAVs (or on drones) is relatively a new area, it remains as a more challenging problem to detect objectsin aerial images. There are several reasons for that including: (i)the lack of large drone datasets including large object variance,(ii) the large orientation and scale variance in drone imageswhen compared to the ground images, and (iii) the differencein texture and shape features between the ground and the aerialimages. Deep learning based object detection algorithms can beclassified under two main categories: (a) single-stage detectorsand (b) multi-stage detectors. Both single-stage and multi-stagesolutions have their advantages and disadvantages over eachother. However, a technique to combine the good sides of each ofthose solutions could yield even a stronger solution than eachof those solutions individually. In this paper, we propose anensemble network, SyNet, that combines a multi-stage methodwith a single-stage one with the motivation of decreasing thehigh false negative rate of multi-stage detectors and increasingthe quality of the single-stage detector proposals. As buildingblocks, CenterNet and Cascade R-CNN with pretrained featureextractors are utilized along with an ensembling strategy. We re-port the state of the art results obtained by our proposed solutionon two different datasets: namely MS-COCO and visDrone with%52.1mAPIoU=0.75is obtained on MS-COCOval2017datasetand %26.2mAPIoU=0.75is obtained on VisDronetest−set. Ourcode is available at: https://github.com/mertalbaba/SyNet


SyNet: An Ensemble Network for Object Detection in UAV Images


Our daily life is surrounded by textual information. Nowadays, the automatic collection of textual information becomes possible owing to the drastic improvement of scene text detectors and recognizer. The purpose of this paper is to conduct a large-scale survey of co-occurrence between visual objects (such as book and car) and scene texts with a large image dataset and a state-of-the-art scene text detector and recognizer. Especially, we focus on the function of "label" texts, which are attached to objects for detailing the objects. By analyzing co-occurrence between objects and scene texts, it is possible to observe the statistics about the label texts and understand how the scene texts will be useful for recognizing the objects and vice versa.

Label or Message: A Large-Scale Experimental Survey of Texts and Objects Co-Occurrence


The age of social media insists on developing devices that are able to capture and share ones’ moments with high fidelity. Handheld devices such as smartphones with wide-angle cameras have shown the current trend in mobile photography. Although one can take great delight in a wide field of view through modern cameras, nearby objects or faces may be distorted significantly. Recent works have obtained impressive results in this research area, but there is still a tradeoff between image quality and processing time to consider. This work introduces an adaptive polynomial model that automatically selects faces and performs image distortion correction. Since the photos are processed locally, faces are undistorted, and the background is close to the original state. Unlike other content-aware based methods which rely on heavy computing components and high image resolution, our model is suitable for mobile devices to tackle face distortion issue in selfie photos.

An Adaptive Model for Face Distortion Correction


Image-based multiplexed in situ RNA detection makes it possible to map the spatial 
gene expression of hundreds to thousands of genes in parallel, and thus discern at the same time a large numbers of different cell types to better understand tissue
development, heterogeneity, and disease. Fluorescent signals are detected over multiple fluorescent channels and imaging rounds and decoded in order to identify RNA molecules in their morphological context. Here we present a graph-based decoding approach that models the decoding process as a network flow problem jointly optimizing observation likelihoods and distances of signal detections, thus achieving robustness with respect to noise and spatial jitter of the fluorescent signals.

Premium content

Learning Group Activities from Skeletons without Individual Action Labels

Downloads

Next from ICPR 2020

Distortion-Adaptive Grape Bunch Counting for Omnidirectional Images

Similar lecture

Learning Group Activities from Skeletons without Individual Action Labels