Jasper Uijlings and Benjamin Risse
Thu 28 May 2015, 11:45 - 13:45
Informatics Forum (IF-4.31/4.33)

If you have a question about this talk, please contact: Steph Smith (ssmith32)

SPEAKER:  Jasper Uijlings
TALK TITLE:  Situational Object Boundary Detection (CVPR 2015)
TALK ABSTRACT:   Intuitively, the appearance of true object boundaries varies from image to image. Hence the usual monolithic approach of training a single boundary predictor and applying it to all images regardless of their content is bound to be suboptimal. In this talk we therefore propose situational object boundary detection: We first define a variety of situations and train a specialized object boundary detector for each of them using [Dollar and Zitnick 2013]. Then given a test image, we classify it into these situations using its context, which we model by global image appearance. We apply the corresponding situational object boundary detectors, and fuse them based on the classification probabilities. In experiments on ImageNet, Microsoft COCO, and Pascal VOC 2012 segmentation we show that our situational object boundary detection gives significant improvements over a monolithic approach. Additionally, our method substantially outperforms [Hariharan et al. 2011] on semantic contour detection on their SBD dataset.

SPEAKER: Benjamin Risse
TALK TITLE: Imaging and Tracking of Semi-Translucent Larvae Using a Novel Multi-Purpose Set-Up
TALK ABSTRACT: Imaging and tracking the locomotion of semi-translucent animals such as Drosophila larvae or C. elegans has become an integral subject of biological research. However, acquiring contrast rich and high resolution measurements is still challenging. Several imaging modalities have been introduced recently, ranging from optical to non-optical acquisition techniques. We designed a novel imaging technique based on frustrated total internal reflection (FTIR) to obtain high resolution and high contrast movies. This FTIR-based Imaging Method (FIM) is suitable for a wide range of biological applications and a wide range of organisms.

In a next step, we have incorporated genetically encoded markers into open-field locomotion experiments by utilising a two-colour FIM setup (FIM2C) that allows the simultaneous detection of fluorescent markers and infrared signals. This enables the implementation of a large variety of perviously impossible experimental approaches. For example, collisions between wild type and GFP-expressing larvae can now be resolved.