Occlusion Robust Part-aware Object Classification through Part Attention and Redundant Features Suppression

AbstractIn this paper, we propose three methods for door state classification with the goal to improve robot navigation in indoor spaces. These methods were also developed to be used in other areas and applications since they are not limited to door detection as other related works are. Our methods work offline, in low-powered computers as the Jetson Nano, in real-time with the ability to differentiate between open, closed and semi-open doors. We use the 3D object classification, PointNet, real-time semantic segmentation algorithms such as, FastFCN, FC-HarDNet, SegNet and BiSeNet, the object detection algorithm, DetectNet and 2D object classification networks, AlexNet and GoogleNet. We built a 3D and RGB door dataset with images from several indoor environments using a 3D Realsense camera D435. This dataset is freely available online. All methods are analysed taking into account their accuracy and the speed of the algorithm in a low powered computer. We conclude that it is possible to have a door classification algorithm running in real-time on a low-power device.

Download Full-text

Grasp to See—Object Classification Using Flexion Glove with Support Vector Machine

Sensors ◽

10.3390/s21041461 ◽

2021 ◽

Vol 21 (4) ◽

pp. 1461

Author(s):

Shun-Hsin Yu ◽

Jen-Shuo Chang ◽

Chia-Hung Dylan Tsai

Keyword(s):

Support Vector Machine ◽

Daily Life ◽

Object Classification ◽

Target Object ◽

The Other ◽

Support Vector ◽

Wide Range ◽

Dark Space ◽

Flex Sensors ◽

Shape And Size

This paper proposes an object classification method using a flexion glove and machine learning. The classification is performed based on the information obtained from a single grasp on a target object. The flexion glove is developed with five flex sensors mounted on five finger sleeves, and is used for measuring the flexion of individual fingers while grasping an object. Flexion signals are divided into three phases, and they are the phases of picking, holding and releasing, respectively. Grasping features are extracted from the phase of holding for training the support vector machine. Two sets of objects are prepared for the classification test. One is printed-object set and the other is daily-life object set. The printed-object set is for investigating the patterns of grasping with specified shape and size, while the daily-life object set includes nine objects randomly chosen from daily life for demonstrating that the proposed method can be used to identify a wide range of objects. According to the results, the accuracy of the classifications are achieved 95.56% and 88.89% for the sets of printed objects and daily-life objects, respectively. A flexion glove which can perform object classification is successfully developed in this work and is aimed at potential grasp-to-see applications, such as visual impairment aid and recognition in dark space.

Download Full-text

Robust Approach to Supervised Deep Neural Network Training for Real-Time Object Classification in Cluttered Indoor Environment

Applied Sciences ◽

10.3390/app11157148 ◽

2021 ◽

Vol 11 (15) ◽

pp. 7148

Author(s):

Bedada Endale ◽

Abera Tullu ◽

Hayoung Shi ◽

Beom-Soo Kang

Keyword(s):

Neural Network ◽

Deep Learning ◽

Real Time ◽

Network Architecture ◽

Input Data ◽

Deep Neural Network ◽

Data Augmentation ◽

Object Classification ◽

Training Data ◽

Gradient Descent Algorithm

Unmanned aerial vehicles (UAVs) are being widely utilized for various missions: in both civilian and military sectors. Many of these missions demand UAVs to acquire artificial intelligence about the environments they are navigating in. This perception can be realized by training a computing machine to classify objects in the environment. One of the well known machine training approaches is supervised deep learning, which enables a machine to classify objects. However, supervised deep learning comes with huge sacrifice in terms of time and computational resources. Collecting big input data, pre-training processes, such as labeling training data, and the need for a high performance computer for training are some of the challenges that supervised deep learning poses. To address these setbacks, this study proposes mission specific input data augmentation techniques and the design of light-weight deep neural network architecture that is capable of real-time object classification. Semi-direct visual odometry (SVO) data of augmented images are used to train the network for object classification. Ten classes of 10,000 different images in each class were used as input data where 80% were for training the network and the remaining 20% were used for network validation. For the optimization of the designed deep neural network, a sequential gradient descent algorithm was implemented. This algorithm has the advantage of handling redundancy in the data more efficiently than other algorithms.

Download Full-text