Search

With the advent of automated species identification and automated trait analysis, researchers need large amounts of training data to improve methods and to gain more insights into the plasticity of traits. Herbaria all over the world have invested large amounts of money and time in collecting samples of plants. However, recent studies have shown that today's methods are not sufficiently able to generalize information acquired on traditional herbarium samples and to transfer them to living samples. This finding stimulated our research into a digital herbarium that fulfils the requirements of traditional collections, but also preserves an individual plant in its living state. Utilizing the various sensors of modern mobile devices, we developed an multi-platform app that seamlessly allows acquiring a digital herbarium sheet. The user initially decides about the life form of a plant to collect and is then being directed through an adaptive serious of image perspectives to be acquired. Completed observations are uploaded to the projects server and become part of a botanists work queue, who is carefully checking and correcting observation details and eventually approving or rejecting it. The observation's status is synced back to the user's device along with detailed information on potential problems discovered by the expert in order motivate the user and to improve observation quality. Approved observations are transferred to conservations bodies. We will demonstrate the end-to-end scenario of capturing an observation, reviewing its correctness, and propagating feedback back to the user. Through Flora Capture, users acquired more than 5,200 observations of more than 500 species already., ICEI 2018 : 10th International Conference on Ecological Informatics- Translating Ecological Data into Knowledge and Decisions in a Rapidly Changing World

Steady improvements of image description methods induced a growing interest in image-based plant species classification, a task vital to the study of biodiversity and ecological sensitivity. Various techniques have been proposed for general object classification over the past years and several of them have already been studied for plant species classification. However, results of these studies are selective in the evaluated steps of a classification pipeline, in the utilized datasets for evaluation, and in the compared baseline methods. No study is available that evaluates the main competing methods for building an image representation on the same datasets allowing for generalized findings regarding flower-based plant species classification. The aim of this paper is to comparatively evaluate methods, method combinations, and their parameters towards classification accuracy. The investigated methods span from detection, extraction, fusion, pooling, to encoding of local features for quantifying shape and color information of flower images. We selected the flower image datasets Oxford Flower 17 and Oxford Flower 102 as well as our own Jena Flower 30 dataset for our experiments. Findings show large differences among the various studied techniques and that their wisely chosen orchestration allows for high accuracies in species classification. We further found that true local feature detectors in combination with advanced encoding methods yield higher classification results at lower computational costs compared to commonly used dense sampling and spatial pooling methods. Color was found to be an indispensable feature for high classification results, especially while preserving spatial correspondence to gray-level features. In result, our study provides a comprehensive overview of competing techniques and the implications of their main parameters for flower-based plant species classification.