A1 Vertaisarvioitu alkuperäisartikkeli tieteellisessä lehdessä
COMPARING CNN-BASED OBJECT DETECTORS ON TWO NOVEL MARITIME DATASETS
Tekijät: Soloviev V, Farahnakian F, Zelioli L, Iancu B, Lilius J, Heikkonen J
Julkaisuvuosi: 2020
Journal: IEEE International Conference on Multimedia and Expo Workshops
Tietokannassa oleva lehden nimi: 2020 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO WORKSHOPS (ICMEW)
Lehden akronyymi: IEEE INT CONF MULTI
Sivujen määrä: 6
ISSN: 2330-7927
Tiivistelmä
Vessel detection studies conducted on inshore and offshore maritime images are scarce, due to a limited availability of domain-specific datasets. We addressed this need collecting two datasets in the Finnish Archipelago. They consist of images of maritime vessels engaged in various operating scenarios, climatic conditions and lighting environments. Vessel instances were precisely annotated in both datasets. We evaluated the out-of-the-box performance of three state-of-the-art CNN-based object detection algorithms (Faster R-CNN [1], R-FCN [2] and SSD [3]) on these datasets and compared them in terms of accuracy and run-time. The algorithms were previously trained on the COCO dataset [4]. We explore their performance based on different feature extractors. Furthermore, we investigate the effect of the object size on the algorithm performance. For this purpose, we group all objects in each image into three categories (small, medium and large) according to the number of occupied pixels in the annotated bounding box. Experiments show that Faster R-CNN with ResNet101 as feature extractor outperforms the other algorithms.
Vessel detection studies conducted on inshore and offshore maritime images are scarce, due to a limited availability of domain-specific datasets. We addressed this need collecting two datasets in the Finnish Archipelago. They consist of images of maritime vessels engaged in various operating scenarios, climatic conditions and lighting environments. Vessel instances were precisely annotated in both datasets. We evaluated the out-of-the-box performance of three state-of-the-art CNN-based object detection algorithms (Faster R-CNN [1], R-FCN [2] and SSD [3]) on these datasets and compared them in terms of accuracy and run-time. The algorithms were previously trained on the COCO dataset [4]. We explore their performance based on different feature extractors. Furthermore, we investigate the effect of the object size on the algorithm performance. For this purpose, we group all objects in each image into three categories (small, medium and large) according to the number of occupied pixels in the annotated bounding box. Experiments show that Faster R-CNN with ResNet101 as feature extractor outperforms the other algorithms.
Turun yliopiston Tutkimustietojärjestelmän portaali