ANOMALY DETECTION(2022–)

Nowadays, remote anomaly detection has become more popular due to the increase in the number of surveillance cameras. However, these surveillance systems are still not timely and require manual labour. Therefore, it is necessary that we leverage the power of computer vision to automatically detect anomalies in videos. The purpose of this problem is to find out a model to accurately identify the start and the end points of an anomalous event.

UIT-Anomaly: A Modern Vietnamese Video Dataset for Anomaly Detection (2022)

VNAnomaly: A Novel Vietnam Surveillance Video Dataset for Anomaly Detection (2022)

IMAGE CAPTIONING(2022–)

Automatically generating captions for images is an exciting subject in computer vision and the natural language processing field. In spite of the precision that recent research has achieved, training an AI model for imitating this unique human ability still has many challenges. In recent years, the common approach for the image captioning problem is based on encoder-decoder architecture like the machine-translation problem: the encoder is a CNN architecture used for extracting visual signals; the decoder is an RNN architecture to predict the possible captions for an according image based on output from the encoder. Inside, feature extraction problem in image captioning has two main approaches: grid features and region features, which one is more effective are now still being discussed.

EAES: Effective Augmented Embedding Spaces for Text-based Image Captioning(2022)

An Augmented Embedding Spaces approach for Text-based Image Captioning(2022)

DOCUMENT IMAGES UNDERSTANDING(2019–)

The COVID19 pandemic has been changing our lives, which requires us to have a proactive approach toward accessing future technologies for manufacturing processes. With digital transformation, paper documents are also gradually converted and replaced by electronic documents for storage on the Cloud Storage, convenient for accessing and searching. The paper documents are stored in images or PDF files format depending on each organization, which leads to many challenges to extract necessary information. This requires a good enough detector model as the foundation for extracting information tasks. The problem’s input is a document image with objects on a possible page: Caption, Table, Figure, and Formula. The output is an image containing the position of the objects expressed by bounding boxes and their labels

Parsing Digitized Vietnamese Paper Documents (2021)

MC-OCR Challenge 2021: Deep Learning Approach for Vietnamese Receipts OCR (2021)

Ensemble of deep object detectors for page object detection (2018)

TRAFFIC ANALYSIS FROM AERIAL IMAGES (2020-)

Unmanned aircraft systems or drones enable us to record or capture many scenes from the bird’s-eye view and they have been fast deployed to a wide range of practical domains, i.e., agriculture, aerial photography, fast delivery and surveillance. Object detection task is one of the core steps in understanding videos collected from the drones. However, this task is very challenging due to the unconstrained viewpoints and low resolution of captured videos. While deep-learning modern object detectors have recently achieved great success in general benchmarks, i.e., PASCAL-VOC and MS-COCO, the robustness of these detectors on aerial images captured by drones is not well studied.

Vehicle Detection at Night Time(2020)

The effects of super-resolution on object detection performance in an aerial image(2020)

Data Augmentation Analysis in Vehicle Detection from Aerial Videos(2020)

Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches(2020)