Instance-level Object Recognition Using Deep Temporal Coherence

Miguel Lagunes Fortiz, Dima Damen, Walterio Mayol-Cuevas

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

148 Downloads (Pure)


In this paper we design and evaluate methods for exploiting temporal coherence present in video data for the task of instance object recognition. First, we evaluate the performance and generalisation capabilities of a Convolutional Neural Network for learning individual objects from multiple viewpoints coming from a video sequence. Then, we exploit the assumption that on video data the same object remains present over a number of consecutive frames. A-priori knowing such number of consecutive frames is a difficult task however, specially for mobile agents interacting with objects in front of them. Thus, we evaluate the use of temporal filters such as Cumulative Moving Average and a machine learning approach using Recurrent Neural Networks for this task. We also show that by exploiting temporal coherence, models trained with a few data points perform comparably to when the whole dataset is available.
Original languageEnglish
Title of host publicationAdvances in Visual Computing
Subtitle of host publication13th International Symposium, ISVC 2018, Las Vegas, NV, USA, November 19 – 21, 2018, Proceedings
PublisherSpringer, Cham
Number of pages12
ISBN (Electronic)9783030038014
ISBN (Print)9783030038007
Publication statusPublished - 8 Nov 2018
Event International Symposium on Visual Computing - Las Vegas, Nevada, Las Vegas, United States
Duration: 19 Nov 201821 Nov 2018

Publication series

NameLecture Notes in Computer Science
ISSN (Print)0302-9743


Conference International Symposium on Visual Computing
Abbreviated titleiscv 2018
Country/TerritoryUnited States
CityLas Vegas
Internet address


  • temporal modeling
  • Deep Learning


Dive into the research topics of 'Instance-level Object Recognition Using Deep Temporal Coherence'. Together they form a unique fingerprint.

Cite this