Labour Disruption Information: yorku.ca/disruption-operations
Close mobile menu

Lassonde Graduate Students Win Six Top Prizes at 2022 AOLS Poster Competition



In early March, the Association of Ontario Land Surveyors (AOLS) held an Annual General Meeting. One of the many events that were hosted was a Graduate Student Poster Session organized by The Association of Ontario Land Surveyors Educational Foundation. Graduate students from the Geomatics program in the Department of Earth and Space Science and Engineering at Lassonde School of Engineering did exceptionally well, taking home all but one of the seven prizes.

Lassonde students took the 1st, 2nd, 3rd, 4th and 5th places in the 2022 poster competition with prizes ranging in value from $500 – $2,000.

The posters were assessed on the content, innovation, clarity, layout, aesthetics and overall presentation by a committee of judges composed of geomatics industry and academia professionals who are members of AOLS.

This success follows last year’s impressive achievement where Lassonde graduate students took home four top prizes.

Learn more about the winning student projects below:

First Place ($2000): Mahya Jodeiri Rad  

Semantic Segmentation of Images using Active Reinforcement Learning 
Semantic segmentation is the process where each pixel on an image is assigned to a specific predefined class. The main challenge with utilizing AI-supervised learning methods is the need for huge, labelled datasets for training. Using a reinforced active learning method, the labelling effort has been reduced while increasing the performance of detecting the critical underrepresented classes. 
Supervised by Professor Costas Armenakis 


Second Place Tie ($1500): Nacer Naciri 

Single-epoch, centimetre-level GNSS point positioning 
Precise Point Positioning (PPP) is a technique that relies on continuous tracking of GNSS signals to achieve centimetre-level positioning over tens of minutes. This work focuses on achieving the same centimetre-level positioning instantaneously, i.e., within one epoch of processing. Results show that such performance is possible with PPP, as 86% of epochs are shown to have a horizontal error of less than 2.5 cm. 
Supervised by Professor Sunil Bisnath 


Second Place Tie ($1500): Ding Yi (pictured in photo) and Sihan Yang 

Resilient smartphone positioning using native sensors in real-world driving scenarios
This research presents a sensor fusion technique utilizing the GNSS receivers and inertial measurement units (IMUs) native to smartphones. Vehicle experiments reveal that the tightly-coupled smartphone PPP/IMU solution is able to mitigate GNSS-only positioning errors significantly by reducing the overall horizontal RMS from 5.2 m to 1.1 m, and more than 80% positioning improvement can be observed with less than 4 visible satellites, showing great potential in providing robust and accurate positioning, velocity and timing for autonomous vehicles, and other mass-market applications. 
Supervised by Professor Sunil Bisnath 


Third Place ($1000): Shamil Samigulin 

Multi-constellation and multi-frequency GNSS signal acquisition strategy for real-time FPGA implementation for applications in GNSS-reflectometry
The goal of this work is to implement the GNSS signal acquisition algorithm on a field-programmable gate array (FPGA) for real-time GNSS-R applications, whilst at the same time minimizing processing time and the hardware resources, therefore, providing a tangible improvement to that of currently existing methods. The overall objective of the work is to build a reliable GNSS-Reflectometry system that is capable of soil moisture content retrieval from the available GNSS signals. The system would be flown on a UAV for field operations and testing in the near future. 
Supervised by Professor Sunil Bisnath and Professor Regina Lee 


Fourth Place Tie ($750): Evangelos Bousias Alexakis  

Building Extraction from Imagery using CNN and Attention-based Architectures 
This work investigates the benefits of incorporating an attention mechanism into a Convolutional Neural Network (CNN) encoder-decoder architecture for end-to-end semantic segmentation of building footprints from images. Attention and self-attention mechanisms attribute more weight to the most relevant input parts of the images allowing them to interact with each other and help the deep learning model to capture rich contextual dependencies between image features. 
Supervised by Professor Costas Armenakis


Fifth Place ($500): Sogand Talebi  

Design and Manufacturing of a GNSS Remote Sensing Payload for Commercial UAVs 
Over the past decade, Unmanned Aerial Vehicle (UAV) technology has become more prevalent in remote sensing, including in land surveying. This poster discusses some of the main design and manufacturing considerations to build a payload that can be seamlessly integrated on a commercial UAV – in this case, the DJI Matrice 300 RTK. 
Supervised by Professor Sunil Bisnath