Amirabolfazl Suratgar

Pooya Janani, Amirabolfazl Suratgar, Afshin Taghvaeipour

This paper proposes a hybrid fusion-based deep learning approach based on two different modalities, audio and video, to improve human activity recognition and violence detection in public places. To take advantage of audiovisual fusion, late fusion, intermediate fusion, and hybrid fusion-based deep learning (HFBDL) are used and compared. Since the objective is to detect and recognize human violence in public places, Real-life violence situation (RLVS) dataset is expanded and used. Simulating results of HFBDL show 96.67\% accuracy on validation data, which is more accurate than the other state-of-the-art methods on this dataset. To showcase our model's ability in real-world scenarios, another dataset of 54 sounded videos of both violent and non-violent situations was recorded. The model could successfully detect 52 out of 54 videos correctly. The proposed method shows a promising performance on real scenarios. Thus, it can be used for human action recognition and violence detection in public places for security purposes.

Milad Behvandi, Seyed Jalil Havakhah, Mohsen Laleh Parvar et al.

In this paper, a legless capsule robot (capsubot) comprised of a sealed external body, an internal body, and a rotational actuator is proposed. The aim of this robot is to move in a two-dimensional viscous environment. After the robot is designed and modeled, a four-stage angular velocity profile is proposed to move the robot in one direction. In addition, a camera is used to obtain the position of the robot in the environment while the orientation and the velocity of the robot is estimated with Kalman filter by fusing the IMU, gyro, and the magnetometer sensors. Furthermore, to control the robot a state feedback control is implemented. Finally, experimental results are provided to demonstrate the performance of the robots and the proposed algorithms.

Saba Heidari Gheshlaghi, Abolfazl Madani, AmirAbolfazl Suratgar et al.

Magnetic resonance images (MRI) play an important role in supporting and substituting clinical information in the diagnosis of multiple sclerosis (MS) disease by presenting lesion in brain MR images. In this paper, an algorithm for MS lesion segmentation from Brain MR Images has been presented. We revisit the modification of properties of fuzzy -c means algorithms and the canny edge detection. By changing and reformed fuzzy c-means clustering algorithms, and applying canny contraction principle, a relationship between MS lesions and edge detection is established. For the special case of FCM, we derive a sufficient condition and clustering parameters, allowing identification of them as (local) minima of the objective function.