From 9e26e4ce6450e08688ac4921b9ff009c2c528a0d Mon Sep 17 00:00:00 2001 From: Jim Martens Date: Wed, 25 Sep 2019 12:48:54 +0200 Subject: [PATCH] Written abstract Signed-off-by: Jim Martens --- abstract.tex | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/abstract.tex b/abstract.tex index 61ca32c..62090f4 100644 --- a/abstract.tex +++ b/abstract.tex @@ -1,4 +1,16 @@ \clearpage \section*{Abstract} -Ich bin ein Abstract +Object detection in open set conditions is an important part of +applying object detection to real world data sets. The key question +is how to identify unknown or novel data. This thesis compares +vanilla SSD and SSD with dropout sampling on the MS COCO data set. +The hyper parameters confidence threshold, entropy threshold, +usage of non-maximum suppression, and usage of dropout layers during +testing are varied to create different scenarios. Both results from macro and +micro averaging are presented. + +Results show a significant improvement of the object detection performance +with a higher confidence threshold. The usage of non-maximum suppression +improves performance as well. Usage of dropout layers during training does +not provide a better result, a lower keep ratio decreases the performance.