Images Of Herbarium Specimens Research Articles

A new species of Pitcairnia (Bromeliaceae) with nocturnal anthesis from Oaxaca, Mexico

Background: Pitcairnia is the third largest bromeliad genus in Mexico with 57 species and ca. 82.5 % endemism, and is characterized mostly by zygomorphic, ornithophilous flowers of diurnal anthesis in several hues of red, orange, or yellow, whereas white or green flowers are uncommon. Moreover, there are just a few species with nocturnal anthesis in Megamexico. One of them is documented and discussed here. Questions: Is the Pitcairnia with campanulate, fragrant flowers of nocturnal anthesis from Oaxaca different from other species with similar characters native to Mexico and the Neotropics and what is its conservation status? Studied species: Pitcairnia abundans, P. lokischmidtiae, P. zapoteca. Study site and dates: Oaxaca, Mexico, 2016-2024. Methods: Live plants, herbarium material, protologues, and high-resolution images of herbarium specimens, both of the new taxon and of morphologically similar taxa, were studied and compared against each other. The conservation status of the new species was assessed using the IUCN criteria. Results: Pitcairnia zapoteca is similar but different at the species level from P. lokischmidtiae from Jalisco; both share campanulate flowers of nocturnal anthesis and a musky or frutal fragances. The conservation status of the new taxon is assessed as DD. Conclusions: Pitcairnia zapoteca is a new species, only known from the type locality in the state of Oaxaca and only one of three Mexican species with campanulate, white, yellowish or greenish flowers of nocturnal anthesis. The assessment the true conservation status of the new taxon requires detailed exploration of the SE portion of Oaxaca.

Development of an Automated Label Data Entry System from Herbarium Specimen Images at Hyogo Herbarium (HYO)

We would like to introduce our recently developed systems for taking images of herbarium specimens and for the automatic extraction of data from specimen labels at the Herbarium of the Museum of Nature and Human Activities, Hyogo, Japan (HYO). Firstly, we designed a low-cost, but high-quality specimen imaging system for non-professional photographers to obtain images rapidly (Takano et al. 2019). Our system uses a mass-produced, mirrorless single-lens reflex (SLR) camera (SONY ILCE6300) with a zoom lens (Samyang Optics SYIO35AF-E35 mm F/2.8). We made a photo stand by ourselves to reduce costs. In addition, we have adopted an LED (light-emitting diode) lighting system with high color rendering. This imaging system has been introduced, with some improvements or adjustments for available space, to various herbaria in Japan (e.g., University of Tokyo (TI), Kyoto University (KYO)), contributing to the digitization of herbarium specimens across Japan. Next, we developed a system to extract label information from specimen images. The specimen image was uploaded to Google OCR and data were extracted in the form of text. Uploading the whole specimen image decreased the reading accuracy of the software because the plant images behaved as OCR (Optical Character Reader) noise. Therefore, the label part was cut out from the whole specimen image by using D-Lib*1 and uploaded to tesseract OCR*2 for OCR extraction of the label information (Aoki 2019, Takano et al. 2020). When installing this system for HYO, we designed it as an application accessible externally via the internet, which proved very useful during the coronavirus pandemic: part-time workers checked and conducted label data input from home. Finally, we decided to develop a system that would automatically label the text data extracted by OCR and input them into the appropriate cells of the database. Even though the text data could be extracted from specimen images, it needed a human to input them into the database. Therefore, we adopted Named Entity Recognition (NER), a system that extracts named entities such as place names, identifying proper nouns from unstructured text data. It enables information recorded in herbarium specimens to be tagged as named entities. We tried text matching at first, but the result was not satisfactory, so we started to use machine learning instead. We compared three natural language libraries for Japanese: BERT (Bidirectional Encoder Representations from Transformers), Albert (A Lite version of BERT), and SpaCy. Despite BERT and SpaCy returning similarly high f-scores (indicating good performance), we decided to use SpaCy because it runs better on ordinary PCs or servers. With sufficient machine learning after the creation of a text corpus (a specialised dataset) specific to labels on herbarium specimens, we successfully developed the application. The project files are available on GitHub*3 (Takano et al. 2024). We then examined whether this system could be applied to non-plant specimen images, i.e., fishes or birds, and found that it could efficiently extract data. Therefore, we decided to publicize this system on the cloud server and share it with other natural history museums in Japan*4. Curators can obtain a unique ID and password and upload specimen images from their collection to extract label data. The digitization of natural history collections in Japan has been long behind other countries, and this system will help to accelerate it. The system mentioned above is specialized for the natural history collections of Japan, but we believe it is possible to build similar programs in other countries, and we hope our experience will contribute to the mobilization of the world’s natural history collections.

Revolutionizing Precision Agriculture: A Comprehensive Review of Innovative Technologies and Application in Digital Farming

Precision agriculture has witnessed a revolution propelled by innovative technologies and applications in digital farming. This comprehensive review explores the transformative impact of these advancements on agricultural practices. Cutting-edge technologies such as agricultural robotics, mechatronic platforms, and autonomous systems are examined for their role in enhancing efficiency and productivity in farming operations. The integration of low-cost location sensing subsystems and adaptive systems further optimizes precision agriculture, enabling real-time decision-making and resource management. Deep learning techniques, specifically in plant disease detection and leaf segmentation from digitized herbarium specimen images, demonstrate remarkable potential for disease management and crop monitoring. Additionally, the design and simulation of robotic arms using machine vision technology represent significant strides in automating tasks within agricultural settings. Through a critical analysis of these technologies and their applications, this review underscores the transformative power of digital farming in revolutionizing precision agriculture. It highlights the potential to address key challenges in sustainable agriculture, improve resource utilization, and meet the growing demands for food production in a rapidly evolving agricultural landscape.

A novel automated label data extraction and data base generation system from herbarium specimen images using OCR and NER

Digital extraction of label data from natural history specimens along with more efficient procedures of data entry and processing is essential for improving documentation and global information availability. Herbaria have made great advances in this direction lately. In this study, using optical character recognition (OCR) and named entity recognition (NER) techniques, we have been able to make further advancements towards fully automatic extraction of label data from herbarium specimen images. This system can be developed and run on a consumer grade desktop computer with standard specifications, and can also be applied to extracting label data from diverse kinds of natural history specimens, such as those in entomological collections. This system can facilitate the digitization and publication of natural history museum specimens around the world.

The Role of the CLIP Model in Analysing Herbarium Specimen Images

The number of openly-accessible digital plant specimen images is growing tremendously and available through data aggregators: Global Biodiversity Information Facility (GBIF) contains 43.2 million images, and Intergrated Digitized Biocollections (iDigBio) contains 32.4 million images (Accessed on 29.06.2023). All these images contain great ecological (morphological, phenological, taxonomic etc.) information, which has the potential to facilitate the conduct of large-scale analyses. However, extracting this information from these images and making it available to analysis tools remains challenging and requires more advanced computer vision algorithms. With the latest advancements in the natural language processing field, it is becoming possible to analyse images with text prompts. For example, with the Contrastive Language-Image Pre-Training (CLIP) model, which was trained on 400 million image-text pairs, it is feasible to classify day-to-day life images by providing different text prompts and an image as an input to the model, then the model can predict the most suitable text prompt for the input image. We explored the feasibility of using the CLIP model to analyse digital plant specimen images. A particular focus of this study was on the generation of appropriate text prompts. This is important as the prompt has a large influence on the results of the model. We experimented with three different methods: a) automatic text prompt based on metadata of the specific image or other datasets, b) automatic generic text prompt of the image (describing what is in the image) and c) manual text prompt by annotating the image. We investigated the suitability of these prompts with an experiment, where we tested whether the CLIP model could recognize a herbarium specimen image using digital plant specimen images and semantically disparate text prompts. Our ultimate goal is to filter the digital plant specimen images based on the availability of intact leaves and measurement scale to reduce the number of specimens that reach the downstream pipeline, for instance, the segmentation task for the leaf trait extraction process. To achieve the goal, we are fine-tuning the CLIP model with a dataset of around 20,000 digital plant specimen image-text prompt pairs, where the text prompts were generated using different datasets, metadata and generic text prompt methods. Since the text prompts can be created automatically, it is possible to eradicate the laborious manual annotating process. In conclusion, we present our experimental testing of the CLIP model on digital plant specimen images with varied settings and how the CLIP model can act as a potential filtering tool. In future, we plan to investigate the possibility of using text prompts to do the instance segmentation to extract leaf trait information using Large Language Models (LLMs).

Extracting Masks from Herbarium Specimen Images Based on Object Detection and Image Segmentation Techniques

Extracting Masks from Herbarium Specimen Images Based on Object Detection and Image Segmentation Techniques

Bamboo Diversity in the Maluku Islands, Indonesia

There is little documentation of bamboo species diversity and distribution in the Maluku Islands of Indonesia. This study aimed to provide information on the diversity and distribution of Maluku’s bamboo species. The research was conducted in March‒April 2021 at the Herbarium Bogoriense (BO), National Research and Innovation Agency. Herbarium specimens of bamboo in BO collected from the Maluku Islands (Provinces of Maluku and North Maluku) were studied alongside online scanned images of herbarium specimens. Bamboo species were re-identified and validated. A literature study was conducted to enrich the data. A distribution map of bamboo species in the Maluku Islands was compiled. Data were analyzed and presented descriptively. We built an identification key for the bamboo species in the Maluku Islands. There are 16 species and eight genera of bamboo in the Maluku Islands: Bambusa glaucophylla, B. maculata, B. multiplex, B. spinosa, B. tuldoides, B. vulgaris, Dendrocalamus asper, Gigantochloa atter, G. robusta, Neololeba amahussana, N. atra, Phyllostachys aurea, Racemobambos ceramica, Schizostachyum brachycladum, S. lima, and Thyrsostachys siamensis. These bamboo species were found across Maluku Province (Buru, Ambon, Seram, Kei Kecil, Kei Besar, Yamdena, and Tanimbar) and North Maluku Province (Ternate, Tidore, Halmahera, Obi, and Sula) at an altitude of 1‒1100 m above sea level in primary and secondary forests, limestone, peatland edges, city parks, and gardens.

Morphometric analysis of the Daphne kiusiana complex (Thymelaeaceae) using digitized herbarium specimens

Daphne kiusiana is an evergreen shrub with dense head-like umbels of white flowers distributed in southern Korea, Japan, China, and Taiwan. Plants in China and Taiwan are recognized as var. atrocaulis by having a dark purple stem, elliptic leaves, and persistent bracts. Recently, plants on Jejudo Island were segregated as a separate species, D. jejudoensis, given their elliptic leaves with an acuminate apex, a long hypanthium and sepals, and a glabrous hypanthium. Morphological variations of three closely related taxa, the D. kiusiana complex, were investigated across the distributional range to clarify the taxonomic delimitation of members of the complex. Twelve characters of the leaf and flower were measured from digitized herbarium specimens using the image analysis program ImageJ and were included in a morphometric analysis, the results of which indicate that the level of variation in the characters is very high. The results of a principal component analysis weakly separated D. jejudoensis from D. kiusiana according to their floral characteristics, such as a longer, glabrous hypanthium, and larger sepals. However, some individuals of D. kiusiana, particularly those from Bigeumdo Island, were included in D. jejudoensis. Recognition of D. kiusiana var. atrocaulis based on the leaf shape was not supported in the analysis, and D. jejudoensis may be recognized as a variety of D. kiusiana. Our morphometric analysis shows that digitized images of herbarium specimens could be useful and an additional method by which to investigate more diverse specimens.

Endemic plants of Java Island, Indonesia: a dataset.

Java is the most populous island in the world. This high population and the extensive economic activities have significantly reduced the forest areas of the Island and have greatly increased the pressure on its plant diversity. Compared to those with a wide distribution, endemic plants with a narrow geographic range are more vulnerable to anthropogenic threats and environmental changes. As species lists are essential for knowledge of species diversity in areas with strong anthropogenic pressure, here we present a dataset of endemic plants of Java Island. The initial species list was manually extracted from the Plant of the World Online (POWO). Each species on the list was then confirmed for its endemism by checking its current distribution using peer-reviewed publications, online plant databases and herbarium specimen images stored on the Global Biodiversity Information Facility (GBIF). The dataset contains 652 species in 279 genera and 85 families. The family with the highest number of endemic species is Orchidaceae (142 species), followed by Rubiaceae (57 species), Acanthaceae (40 species), Apocynaceae (35 species) and Lauraceae (29 species). The growth form of the species is mostly trees (22.6%), followed by herbs (19.2%), epiphytes (16%), shrubs (12.4%), vines (11%) and geophytes (9.4%). Most of the species (89.7%) have not yet been assessed for their conservation status according to the IUCN Red List Categories and Criteria. There are only 55 species (8.3%) that have been conserved within ex situ collections. Furthermore, most of the species (79.8%) are not listed on the CITES appendices and there are only four species (0.6%) protected by national law. Our contribution provides the first online list of accepted scientific names of Javan endemic plants species, together with all their synonyms. New to the dataset are: i) provision of local names of the species (if available), ii) the classification of species under eleven growth forms (tree, shrub, herb, annual, graminoid, geophyte, fern, vines, hydrophyte, parasite and epiphyte), iii) assignation of the extinction risk of species according to the IUCN Red List, iv) ex situ collection status of species and information on the protection status of the species according to (v) CITES and (vi) the national law of Indonesia.

The effectiveness and limitations of digital images for taxonomic research

AbstractIn response to the worldwide coronavirus outbreak, which effectively shut down fieldwork, laboratory and herbarium‐based studies, an evaluation was made of the effectiveness and limitations of undertaking a virtual taxonomic study using only online herbarium specimen resources related to the genus Madhuca (Sapotaceae) for the Flora of Singapore. The study demonstrated the immense value of digital images to basic taxonomic research but also found that diagnostic micro‐morphological characters, often critical in defining species boundaries, cannot be seen in many digital images, even at high resolution. Several recommendations are made on how to maximise the utility of online herbarium specimen images to help facilitate future taxonomic research, though it is clear that physical access to herbarium specimens remains essential.

Automatic Identification of Ivorian Plants from Herbarium Specimens using Deep Learning

— Plant identification is most often based on visual observations by botanists and systematists. Deep learning has become a tool that provides an alternative to automatic plant identification. Our study consists in implementing a method for plant recognition from herbarium specimens using deep learning classification methods. These methods were evaluated on the dataset of ten plant families from the national herbarium of Côte d'Ivoire. The proposed work uses CNN architectures such as DensNet-121, InceptionV3, VGG19, MobileNet, and ResNet101. The dataset contains 7543 images of herbarium specimens. The database is structured in three parts: training, testing, and validation. The accuracies obtained for the first scenario without preprocessing of herbarium specimen images are 76.94% for MobileNet, 77.77% for VGG19, and 77.96% for InceptionV3, 80.41% for ResNet101, and 83.47% for DensNet-121, respectively. The best performance was obtained with DensNet-121 with 83.47%. In the second scenario with preprocessing of herbarium specimens, the accuracies obtained were 82.80% for InceptionV3, 84.40% for VGG19, 85.53% for MobileNet, and 85.80% for ResNet101. The best accuracy was obtained with ResNet121 with 85.80%. From the analysis obtained, the results show that ResNet101 gives the best accuracy compared to the other architectures. In particular, the data preprocessing improves the prediction results, of the Convolutional Neural Network algorithms. Keywords— Deep learning, Herbarium specimens, image preprocessing, Convolutional Neural Network, Classification.

A deep learning-based approach for detecting plant organs from digitized herbarium specimen images

Herbarium specimens are excellent sources of botanical information to facilitate understanding and monitoring the evolution of plants and their effects on global climate change. Globally, many herbaria have undertaken digitization projects of herbarium specimens to preserve them and make them accessible in online repositories to botanists and ecologists. Automated detection of plant organs such as plant leaves, buds, flowers, and fruits on the digitized herbarium specimen images provides valuable information in various scientific contexts. We developed a deep learning approach based on the refined YOLO-V3 approach to detect plant organs within the digitized herbarium specimen images effectively. The proposed approach combines ResNet and DenseNet architectures to improve feature extraction capabilities. Also, a new scale of feature map is added to the existing scales to address the problem of YOLO-V3's low performance in detecting small plant organs. The experimental results demonstrate that our proposed approach can detect organs of different sizes within different specimens, where the precision and recall reached 94.2% and 95.5%, respectively.

The Herbarium 2021 Half-Earth Challenge Dataset and Machine Learning Competition.

Herbarium sheets present a unique view of the world's botanical history, evolution, and biodiversity. This makes them an all–important data source for botanical research. With the increased digitization of herbaria worldwide and advances in the domain of fine–grained visual classification which can facilitate automatic identification of herbarium specimen images, there are many opportunities for supporting and expanding research in this field. However, existing datasets are either too small, or not diverse enough, in terms of represented taxa, geographic distribution, and imaging protocols. Furthermore, aggregating datasets is difficult as taxa are recognized under a multitude of names and must be aligned to a common reference. We introduce the Herbarium 2021 Half–Earth dataset: the largest and most diverse dataset of herbarium specimen images, to date, for automatic taxon recognition. We also present the results of the Herbarium 2021 Half–Earth challenge, a competition that was part of the Eighth Workshop on Fine-Grained Visual Categorization (FGVC8) and hosted by Kaggle to encourage the development of models to automatically identify taxa from herbarium sheet images.

Harnessing Large-Scale Herbarium Image Datasets Through Representation Learning.

The mobilization of large-scale datasets of specimen images and metadata through herbarium digitization provide a rich environment for the application and development of machine learning techniques. However, limited access to computational resources and uneven progress in digitization, especially for small herbaria, still present barriers to the wide adoption of these new technologies. Using deep learning to extract representations of herbarium specimens useful for a wide variety of applications, so-called “representation learning,” could help remove these barriers. Despite its recent popularity for camera trap and natural world images, representation learning is not yet as popular for herbarium specimen images. We investigated the potential of representation learning with specimen images by building three neural networks using a publicly available dataset of over 2 million specimen images spanning multiple continents and institutions. We compared the extracted representations and tested their performance in application tasks relevant to research carried out with herbarium specimens. We found a triplet network, a type of neural network that learns distances between images, produced representations that transferred the best across all applications investigated. Our results demonstrate that it is possible to learn representations of specimen images useful in different applications, and we identify some further steps that we believe are necessary for representation learning to harness the rich information held in the worlds’ herbaria.

Deep leaf: Mask R-CNN based leaf detection and segmentation from digitized herbarium specimen images

• A modified mask RCNN named Deep Leaf is developed to identify the leaves from the digitized herbarium specimens. • Deep Leaf measures automatically the morphological traits of the extracted leaves. • Deep features are extracted through an improved ResNet50/101, which is chosen as the backbone network of the feature extraction. • We achieved better performance compared with the original mask RCNN algorithm for leaves detection. The generation of morphological traits of plants such as the leaf length, width, perimeter, area, and petiole length are fundamental features of herbarium specimens, thus providing high-quality data to investigate plant responses to ongoing climatic change and plant history evolution. However, the existing measurement methods are primarily associated with manual analysis, which is labor-intensive and inefficient. This paper proposes a deep learning-based approach, called Deep Leaf, for detecting and pixel-wise segmentation of leaves based on the improved state-of-the-art instance segmentation approach, Mask Region Convolutional Neural Network (Mask R-CNN). Deep Leaf can accurately detect each leaf in the herbarium specimen and measure the associated morphological traits. The experimental results indicate that our automated approach can segment the leaves of different families. Compared to manual measurement done by ecologist and botanist experts, the average relative error of leaf length is 4.6%, while the average relative error of leaf width is 5.7%.

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries.

BackgroundMorphological leaf traits are frequently used to quantify, understand and predict plant and vegetation functional diversity and ecology, including environmental and climate change responses. Although morphological leaf traits are easy to measure, their coverage for characterising variation within species and across temporal scales is limited. At the same time, there are about 3100 herbaria worldwide, containing approximately 390 million plant specimens dating from the 16th to 21st century, which can potentially be used to extract morphological leaf traits. Globally, plant specimens are rapidly being digitised and images are made openly available via various biodiversity data platforms, such as iDigBio and GBIF. Based on a pilot study to identify the availability and appropriateness of herbarium specimen images for comprehensive trait data extraction, we developed a spatio-temporal dataset on intraspecific trait variability containing 128,036 morphological leaf trait measurements for seven selected species.New informationAfter scrutinising the metadata of digitised herbarium specimen images available from iDigBio and GBIF (21.9 million and 31.6 million images for Tracheophyta; accessed date December 2020), we identified approximately 10 million images potentially appropriate for our study. From the 10 million images, we selected seven species (Salix bebbiana Sarg., Alnus incana (L.) Moench, Viola canina L., Salix glauca L., Chenopodium album L., Impatiens capensis Meerb. and Solanum dulcamara L.) , which have a simple leaf shape, are well represented in space and time and have high availability of specimens per species. We downloaded 17,383 images. Out of these, we discarded 5779 images due to quality issues. We used the remaining 11,604 images to measure the area, length, width and perimeter on 32,009 individual leaf blades using the semi-automated tool TraitEx. The resulting dataset contains 128,036 trait records.We demonstrate its comparability to trait data measured in natural environments following standard protocols by comparing trait values from the TRY database. We conclude that the herbarium specimens provide valuable information on leaf sizes. The dataset created in our study, by extracting leaf traits from the digitised herbarium specimen images of seven selected species, is a promising opportunity to improve ecological knowledge about the adaptation of size-related leaf traits to environmental changes in space and time.

Automated Extraction of Phenotypic Leaf Traits of Individual Intact Herbarium Leaves from Herbarium Specimen Images Using Deep Learning Based Semantic Segmentation.

With the increase in the digitization efforts of herbarium collections worldwide, dataset repositories such as iDigBio and GBIF now have hundreds of thousands of herbarium sheet images ready for exploration. Although this serves as a new source of plant leaves data, herbarium datasets have an inherent challenge to deal with the sheets containing other non-plant objects such as color charts, barcodes, and labels. Even for the plant part itself, a combination of different overlapping, damaged, and intact individual leaves exist together with other plant organs such as stems and fruits, which increases the complexity of leaf trait extraction and analysis. Focusing on segmentation and trait extraction on individual intact herbarium leaves, this study proposes a pipeline consisting of deep learning semantic segmentation model (DeepLabv3+), connected component analysis, and a single-leaf classifier trained on binary images to automate the extraction of an intact individual leaf with phenotypic traits. The proposed method achieved a higher F1-score for both the in-house dataset (96%) and on a publicly available herbarium dataset (93%) compared to object detection-based approaches including Faster R-CNN and YOLOv5. Furthermore, using the proposed approach, the phenotypic measurements extracted from the segmented individual leaves were closer to the ground truth measurements, which suggests the importance of the segmentation process in handling background noise. Compared to the object detection-based approaches, the proposed method showed a promising direction toward an autonomous tool for the extraction of individual leaves together with their trait data directly from herbarium specimen images.

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries

Comprehensive leaf size traits dataset for seven plant species from digitised herbarium specimen images covering more than two centuries