"GastroView - nowa technologia analizy i przetwarzania obrazu, zwiększająca dokładność oraz skracająca czas diagnostyki zmian chorobowych w obszarze endoskopii kapsułkowej", projekt dofinansowany w ramach Poddziałania 1.1.1 "Badania przemysłowe i prace rozwojowe realizowane przez przedsiębiorstwa" Programu Operacyjnego Inteligentny Rozwój 2014-2020 współfinansowanego ze środków Europejskiego Funduszu Rozwoju Regionalnego.

Angiodysplasia detection algorithm in the GastroView system uses a convolutional neural network as a base classification model. The network is trained on image patches instead of entire images in order to operate in sliding-window mode in test time. The outputs are then aggregated and postprocessed in order to provide detection (full frame classification) and localization outputs.

In this short experiment we will implement simple transfer learning with Keras framework in order to get a fairly reliable classifier with just a little effort and using really little data. For this purpose we will use Resnet-152 model that was trained on ImageNet and finetune it for the task of classifying interiors of 67 different places. The data used in the experiment comes from the MIT Indoor-67 dataset.

CV Lab enables convenient development of computer vision algorithms by means of graphical designing of the processing flow. Writing code with OpenCV might be a time-consuming process. It is often required to compile and run the code in order to see the results of the currently made changes. Especially, when some of the function parameters are to be tuned for establishing the optimal values. Some code also has to be added to provide presentation of the intermediate or final results of the algorithm.