Invited Talks

Invited Talk: Ass. Prof. Anastasia Zakharova (Laboratoire MIA, La Rochelle Université, France)

Title: Practicable Inductive Graph Neural Networks for Moving Objects Detection 

Abstract

Moving Object Segmentation (MOS) is a challenging problem in computer vision, particularly in scenarios with dynamic backgrounds, abrupt lighting changes, shadows, camouflage, and moving cameras. While graph-based methods have shown promising results in MOS, they have mainly relied on transductive learning which assumes access to the entire training and testing data for evaluation. However, this assumption is not realistic in real-world applications where the system needs to handle new data during deployment. In this paper, we propose a novel Graph Inductive Moving Object Segmentation (GraphIMOS) algorithm based on a Graph Neural Network (GNN) architecture. Our approach builds a generic model capable of performing prediction on newly added data frames using the already trained model. GraphIMOS outperforms previous inductive learning methods and is more generic than previous transductive techniques. Our proposed algorithm enables the deployment of graph-based MOS models in real-world applications.

Invited Talk: Prof. Tomasz Kryjak, (AGH University of Krakow, Poland) 

Title: Real-Time Processing of DVS (event camera streams) using GNNs and FPGAs 

Abstract:

Energy-efficient, real-time perception under adverse lighting conditions is critical for embedded computer vision systems and is driving the adoption of advanced neuromorphic sensors known as event cameras (Dynamic Vision Sensors - DVS). DVS uniquely capture only changes in the observed scene, operating independently and asynchronously at the pixel level. This results in low power consumption, high throughput, reduced latency and minimal motion blur, even under challenging lighting conditions, making them ideal for perception algorithms in dynamic environments, such as mobile robotics. This talk will focus mainly on Graph Convolutional Neural Networks designed to process the sparse data generated by DVS, and their implementation for embedded platforms, in particular heterogeneous FPGAs (Field Programmable Gate Arrays). The approach developed in our research group allows asynchronous, continuous and efficient processing of event data while maintaining high classification and detection accuracy.