Master in Computer Vision Barcelona 2020 - Module 6: Video Analysis

Abstract

Week 1 Implementing dataset and evaluation metrics

This week we implemented the evaluation metrics(Intersection over Union (IoU) and mean Average Precision (mAP)) that have been used in all tasks again.
[IoU with noise DB ] [mAP with noise DB ]

Week 2 Object detection with Gaussian model

The main task is to use Gaussian models to model the background pixels of the video sequence to classify them as background / foreground.
[bbox_adaptive]

Week 3 Object Detection and Object Tracking

Object Detection Mask-RCNN framework with ResNet-101 as the backbone
[Mask-RCNN] [Faster-RCNN]

Week 4 Optical Flow and Video Stabilization

[Optical Flow] [Video Stabilization]

Week 5 Multi-target multi-camera tracking

[Result of Multi-target multi-camera tracking through Color Histogram + metric learning]

Final slides and report

Slides of week5
Technical report

Team members:

Member	Email
Yixiong Yang	yangyxwork@163.com
Sanket Biswas	sanket.biswas@e-campus.uab.cat
Gabriela Cordova	gabrielaelizabeth.cordova@e-campus.uab.cat
Marc Oros Casanas	marc.oros@e-campus.uab.cat
Keyao Li	keyao.li@e-campus.uab.cat