This is the official Python and C++ implementation repository for a paper entitled "Track Initialization and Re-Identification for 3D Multi-View Multi-Object Tracking", Information Fusion (http://arxiv.org/abs/2405.18606).

Comparative analysis of DeepOCSORT, StrongSORT, and ByteTrack for robust tracking under heavy occlusion.

Face Under Mask Reconstruction , Image Inpainting, and Object Detection and Segmentation with the Detectron2 library

Unsupervised deep framework that animates a source image using motion extracted from a driving video through learned keypoints, local affine transforms, dense motion fields, and occlusion-aware generation.

This is the official Python and C++ implementation repository for a paper entitled "Visual multi-object tracking with re-identification and occlusion handling using labeled random finite sets", Pattern Recognition (https://arxiv.org/pdf/2407.08872).

This face recognition implementation is capable of recognizing faces with a certain level of occlusion, this includes faces wearing masks. You can also add new users manually by adding a photo in the images folder.

Public toolkit for HOOT: Heavy Occlusions for Object Tracking Benchmark.

An instance segmentation challenge on Basketball images, with a particular focus on occlusion resolution. An opportunity to publish at MMSports @ ACMMM and to win $1000.

Deep Occlusion-Aware Instance Segmentation with Overlapping BiLayers [CVPR 2021]

Noise Aware OSTrack

3D normal map estimation pipeline using low resolution and highly occluded images

Knowledge distillation for masked FER using ResNet-18 in PyTorch.

Behind the Curtain: Learning Occluded Shapes for 3D Object Detection

Master thesis project realized in partnership with Aalto University and Finnish Geospatial Research Institute of Finland. Stereo Camera-LiDAR calibration. Covers: mono and stereo camera calibration, Camera-LIDAR synchronisation, camera-based and LiDAR-based 3D reconstruction, Camera-LiDAR extrinsic calibration, Occlusion handling.