Slovakia

&quot;Veterinary medicine emphasizes accidents caused by toxic plants with domestic animals as an extremely important topic, as the right diagnosis can be crucial for the affected animal. In this work, we propose the classification of toxic ornamental plants, according to nine different categories, using five widely-known CNN architectures, namely: DenseNet, ResNet, VGG16, VGG19 and Xception. The rationale behind it is that the automatic identification of these types of plant can be a useful tool to help in the prevention of those accidents. The authors have carefully curated a database to support the development of this work, collecting images available on the Pinterest website, and also performing some important data pre-processing. This database was also made available as a contribution of this work.
Transfer learning was employed by taking advantage of feature learned from the ImageNet dataset. We also analyzed the heat maps generated by the Layer-wise Relevant Propagation method, which allowed to observe the individual behavior of the best and worst architectures. The best performance was achieved using DenseNet, with an accuracy of 97.67%. That model managed to generalize very well, even to deal with noisy images, which are frequent in photos of decorative environments.&quot;

IWSSIP 2021

Classification of Toxic Ornamental Plants for Domestic Animals using CNN

heat map

densenet architecture

plant identification

data pre processing

toxic plant

layer wise relevance propagation

domestic animal

toxic ornamental plant

convolutional neural network

deep learning

data augmentation

transfer learning


IWSSIP is an international conference that brings together researchers and developers from both academia and industry reporting the latest scientific and theoretical advances to discuss and debate major issues and state-of-the-art aspects on systems, signals and image processing.

###### For the session discussion, each author will be seated at individual breakout tables.  Please find which table they will be at here:
###### [QA table assignments in Gather.town](https://drive.google.com/file/d/1hBHOD_KF7NSsaTcpYcHCZslXJvQVxJQe/view?usp=sharing)

"Veterinary medicine emphasizes accidents caused by toxic plants with domestic animals as an extremely important topic, as the right diagnosis can be crucial for the affected animal. In this work, we propose the classification of toxic ornamental plants, according to nine different categories, using five widely-known CNN architectures, namely: DenseNet, ResNet, VGG16, VGG19 and Xception. The rationale behind it is that the automatic identification of these types of plant can be a useful tool to help in the prevention of those accidents. The authors have carefully curated a database to support the development of this work, collecting images available on the Pinterest website, and also performing some important data pre-processing. This database was also made available as a contribution of this work.
Transfer learning was employed by taking advantage of feature learned from the ImageNet dataset. We also analyzed the heat maps generated by the Layer-wise Relevant Propagation method, which allowed to observe the individual behavior of the best and worst architectures. The best performance was achieved using DenseNet, with an accuracy of 97.67%. That model managed to generalize very well, even to deal with noisy images, which are frequent in photos of decorative environments."

technical paper

"Atrial fibrillation (AF) is a cardiovascular disease identified as one of the main risks
factors for stroke. The majority of strokes due to AF are caused by clots originating in the left atrial appendage (LAA). LAA occlusion is an effective procedure for reducing stroke risk.  Planning the procedure using pre-procedural imaging and analysis has shown benefits. The analysis is commonly done by manually segmenting the appendage on 2D slices. Automatic LAA segmentation methods could save an expert's time and provide insightful 3D visualizations and accurate automatic measurements to aid in medical procedures. Several semi- and fully automatic methods for segmenting the appendage have been proposed. This paper provides a review of automatic LAA segmentation methods on 3D and 4D medical images, including CT, MRI, and echocardiogram images. We classify methods into heuristic and model-based methods, as well as semi- and fully-automatic methods. We summarize and compare the proposed methods, evaluate their effectiveness, present current challenges in the field and approaches to overcome them."

Left Atrial Appendage Segmentation and Analysis in 3D and 4D medical images

This presentation shows the results of our research based on the comparison of approaches of automatic detection and localization of an optic disc and macula using on deep learning. Information about their position helps to diagnose diabetic retinopathy - a disease that causes visual impairment and leads to blindness. Detecting the symptoms of this disease manually is a demanding process, but thanks to advances in computer vision, it can be automated. Compared methods base on Faster R-CNN, YOLO and SSD detectors are used to detect the optic disc and macula in the eye fundus images. We test the methods on three publically available datasets of fundus images - “DRIVE“, “IOSTAR“ and “ODIR“. We created own dataset containing 500 fundus images as a combination of mentioned databases, where the ground truth position of the macula and optic disc were marked by an ophthalmologist.

Detection of optic disc and macula in retinal images using deep learning methods

The motivation for the research in this paper is gathering valuable information to develop a DFT-based steganographic method using an existing DFT-based watermarking technique. Specifically, different image dimensions will be tested to find out how the difference in dimension affects the ability to decode a watermark from an image successfully. The goal is to determine the smallest possible image dimension that still produces successful decoding. Determining that image dimension is critical to examine to develop a steganographic method with the highest capacity possible. Several experiments are conducted to draw conclusions in terms of image quality, detection rate, and robustness. In this research, tests have been conducted in the digital domain exclusively. The results show that image dimensions play a critical role in watermark detection, and further development of the method is needed in order to be able to use images of smaller dimensions. The watermarking method showed robustness for several image attacks and showed enhanced performance when specific filters are applied on images.

Analysis of the DFT-based Watermarking Method for Image Steganography

Current wireless communication in ad-hoc networks requires a so-phisticated approach for the data location, retrieval, and transfer, to use the con-nection availability appropriately. Data are commonly stored in the databases spread across the network or located in each node. This paper deals with the current techniques of data location and retrieval using the index set. The main limitation is the availability and suitability of the index for the particular query request. Our proposed solution is based on the multi-indexing layer, by which the several indexes can be used to a single query in parallel, followed by the merging operation using bitmap index mapping. Thanks to the proposed archi-tecture, data location and efficiency of the whole system can be ensured. The robustness and scalability of the solution can be reached, as well.

Using multi-index database layer in ad-hoc communication networks

This article describes a method for detection of cognitive defects based on eye tracking during reading. The aim of this research is to pursue and extend the experiments conducted in Sweden by introducing Conventional signal theory. The dataset used for experiment was acquired by the authors of the research article Screening for Dyslexia Using Eye Tracking during Reading. The provided data consist of 185 subjects divided into two groups. The first group comprises of 88 low risk (LR) subjects and the second group comprises of 97 high risk (HR) subjects. Our measurements achieved a classification accuracy score 86.164% by classifying the subjects into the correct groups.

Fast Algorithm for Dyslexia Detection

"An unsolved problem in the domain of Internet Accessibility, which is a stream in Human Computer Interface (HCI), is how cognitively challenged or visually impaired users access a service. This paper proposes an assistive technology for disabled users in the form of a Randomized Authentication Gateway (RAG) mechanism corresponding to a challenge-pool. The challenge-pool contains different stand-alone Accessibility Protocols that are suitable for different situations. Holistically,
we believe that the system works the finest when it is used as a single federated interface. The RAG manager picks up random challenges from the challenge-pool and the client, who is looking to access a particular service, is subjected to these challenges. If the client successfully completes them, he is authenticated as a valid user of the intended service. This paper discusses the audio-based, pressure-based and gesture-based mechanisms for seamless authentication of challenged users. We have defined this problem in terms of Accessibility Deficiency and Bot Indicator. The proposed system is only 23% accessibility deficient and covers the use cases of partially blind, completely blind, color blind and users suffering from diseases that affects their comprehension skills, like dyslexia. The research focuses on a keyboard-friendly environment where the fit or challenged users will have complete control over the interface using the one-click functionality. This work is our way to ensure that no restriction is imposed on any user while surfing through the internet."

Secure Web Accessibility Protocol for Challenged Users

In this paper we present our GNU Radio based Software-Defined Radio laboratory configured to leverage the practical benefits provided by vir-tualization for lab installation and operation. In the context of recent social dis-tancing-mandated remote operation requirements the procedure for connecting the laboratory online, while adhering to the security guidelines required by the institutes network policy, is described in detail. This setup, combining the virtu-alized configurations with remote access through Virtual Private Networking allows the staff and students to access the specialized laboratory equipment – the USRP signal processors, located on premises, conveniently from their homes, thus eliminating the need for otherwise necessary lab attendance.

A Remote-Access Software-Defined Radio Virtualized Wireless Communications Laboratory

In this contribution we focus on technical and economic aspects of using battery storage in microgrids for peak shaving. We perform feasibility anal-ysis of large battery storage usage in microgrid systems under the current Slovak legislative windows. We use multiple real-life consumption pro-files from various industry oriented microgrids. The primary usage of bat-tery storage in our models is peak shaving and possible reduction of re-quired reserved capacity. The main objective is to find the equilibrium be-tween technical innovations, acquisition costs and energy cost savings.

Peak shaving in microgrids using battery storage

rosurgeon and quantitatively based on statistical tests. These viewpoints

Comparison of TCP Receiver-based Loss Differentiation Algorithms in Wired/Wireless Networks

"The major contributions of our approach are:

1. Implementation of model with mathematical concepts for analyzing the overall delay results during handover.
2. Verification model is presented with necessary verification procedures involved in the sensitivity analysis including complexity using virtual data plane switches.
"

Premium content

Classification of Toxic Ornamental Plants for Domestic Animals using CNN

Downloads

Next from IWSSIP 2021

Left Atrial Appendage Segmentation and Analysis in 3D and 4D medical images

Similar lecture

Deep learning-based detection of seedling development from indoor to outdoor