|
Applying Deep Learning for Mask Detection to Prevent Airborne Disease Spreading |
|
Kittitham Padungwiang, Anuntapat Anuntachai, Chanin Thongchin(King Mongkut’s Institute of technology Ladkrabang, Thailand) |
| Airborne diseases are very easily infected when people do daily activities such as breathing, talking or coughing because virus or bacteria float in the air. An apparent example is COVID-19 pandemic which currently spread over the world. The most important and simply prevention is to wear surgical mask. However, people are not willing to wear facial mask in public place or they wear mask with wrong instruction, these increase inefficiency of defensive measure and spreading areas. This research studies on mask wearing detection to develop correctness facial mask wearing detection. The research |
|
|
TypicalVietnameseFoodNet: A Vietnamsese Food Image Dataset For Vietnamese Food Classification |
|
Tri Thien Cao, Khoa Van Duong(University of Science, Vietnam Vietnam National University,Ho Chi Minh City, Viet Nam) |
| Vietnamese food classification connects us across our cultures and generations. Food classification is not easy, even with people. The reason is the food’s extreme diversity between dishes and in the middle variations of the dish. So some traditional approaches with hand-crafted features had been used for food recognition. We propose a new dataset called TypicalVietnameseFoodNet and a proposed model with the best performance for our dataset. |
|
|
Prediction of Pedestrian Trajectory based on Long Short-term Memory of Data |
|
Tomoya Ono, Takashi Kanamaru(Kogakuin University, Japan) |
| We predict pedestrian trajectories with LSTM (long short-term memory) networks and CNN (conventional neural networks), and we compare their results. In order to predict sequential data, we use the following two methods: (I) predicting n steps of data with n models, and (II) predicting n steps of data with a model by applying one-step prediction several times. By examining these methods, it was found that the performances of the method I with the LSTM network and the CNN are comparable, and the performance of the method II with the LSTM network is significantly better than that with the CNN. |
|
|
Domain Adaptation for Agricultural Image Recognition and Segmentation Using Category Maps |
|
Kota Takahashi(Akita Prefectural University, Japan), Hirokazu Madokoro(Iwate Prefectural University, Japan), Satoshi Yamamoto, Yo Nishimura, Stephanie Nix(Akita Prefectural University, Japan), Hanwool Woo(The University of Tokyo, Japan), Takashi K. Saito, Kazuhito Sato(Akita Prefectural University, Japan) |
| Recognition accuracy obtained using deep learning drops precipitously when the training data are insufficient. This paper presents a data-expansion method for training of the transfer learning source domain. Experiment results obtained from two open benchmark datasets and our original benchmark dataset demonstrated that our proposed method outperforms the previous method under a guarantee of sufficient accuracy for the synthetic images. |
|
|
Hypergraph based Multi-Agents Representation Learning for Similarity Analysis |
|
Jaeuk Baek(ETRI, Korea), Changeun Lee(Electronics and Telecommunications Research Institute (ETRI), Korea) |
| In this paper, we propose a hypergraph based multi-agents representation learning (HMARL) to obtain agent embedding vectors, which can be used to classify agents in the same region and correlate the collected data of similar properties. To this end, the proposed HMARL transforms the multi-modal data into the same graph structure with nodes and their relations. Then, a hypergraph is constructed to integrate local graphs and a hypergraph random walk is applied to obtain the sequence of adjacent agents, which is used to train the agent embedding vectors. Experiments on public datasets are provided for similarity analysis on agents and their collected data. |
| |