Pl@ntNet-300K is an image dataset aimed at evaluating set-valued classification. It was built from the database of Pl@ntnet citizen observatory and consists of 306146 images, covering 1081 species. We highlight two particular features of the dataset, inherent to the way the images are acquired and to the intrinsic diversity of plants morphology:
i) The dataset exhibits a strong class imbalance, meaning that a few species represent most of the images.
ii) Many species are visually similar, making identification difficult even for the expert eye.
These two characteristics make the present dataset a good candidate for the evaluation of set-valued classification methods and algorithms. Therefore, we recommend two set-valued evaluation metrics associated with the dataset (top-K and average-K) and we provide the results of a baseline approach based on a resnet50 trained with a cross-entropy loss. The full description of the dataset can be found in (to be provided soon).

The scientific publication (NEURIPS 2022) describing the dataset and providing baseline results can be found here: https://openreview.net/forum?id=eLYinD0TtIt

Utilities to load the data and train models with pytorch can be found here: https://github.com/plantnet/PlantNet-300K/

There are two datasets and one table uploaded in this platform under the title "MED117_Medicinal Plant Leaf Dataset & Name Table". A folder is created with title "MED 117 Leaf Species". Inside this two sub folders with titles " Raw leaf image set of medicinal plants_v2" and "Segmented leaf set using UNET segmentation" are created. Raw leaf image set consists of leaf images of 117 medicinal plants found in Assam. All the samples are collected by visiting different (Govt, Public and Private) medicinal gardens situated in different places of Assam and some other general places where they are mostly found. Videos of 10 to 15 seconds duration were taken for two to three leaves of every species on a white background and video recording was done using a SLR Canon Camera. Individual videos were segregated into image frames and thus were able to get around 77,700 jpg image frames from the videos. The Raw leaf image set consists of folders with scientific name and common name within bracket. Second folder with title "Segmented leaf set using UNET segmentation" consists of 115 medicinal plant species with their segmented leaf image samples using UNET segmentation technique. Here two species are excluded from the original dataset due to small unpredictable size of the samples, so total 115 subfolders inside the segmented folder is achieved. Thirdly a table in doc format with title "Medicinal Plant Name Table" is uploaded and it includes Scientific name, Common name and Assamese name of the plants listed in the folders in the same sequence. The whole contribution is absolutely original and new, collected from different sources then processed for segmentation and prepared the table by discussing with taxonomy experts from Botany department of Gauhati University, Guwahati, Assam. India.

Fruit and vegetable plants are vulnerable to diseases that can negatively affect crop yield, causing planters to incur significant losses. These diseases can affect the plants at various stages of growth. Planters must be on constant watch to prevent them early, or infestation can spread and become severe and irrecoverable. There are many types of pest infestations of fruits and vegetables, and identifying them manually for appropriate preventive measures is difficult and time-consuming.This pretrained model can be deployed to identify plant diseases efficiently for carrying out suitable pest control. The training data for the model primarily includes images of leaves of diseased and healthy fruit and vegetable plants. It can classify the multiple categories of plant infestation or healthy plants from the images of the leaves.Using the modelFollow the guide to use the model. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS. Fine-tuning the modelThis model can be fine-tuned using the Train Deep Learning Model tool. Follow the guide to fine-tune this model.Input8 bit, 3-band (RGB) image. Recommended image size is 224 x 224 pixels. Note: Input images should have grey or solid color background with one full leaf per image. OutputClassified image of the leaf with any of the plant disease, healthy leaf, or background classes as in the Plant Leaf Diseases dataset.Applicable geographiesThis model is expected to work well in all regions globally. However, results can vary for images that are statistically dissimilar to training data.Model architectureThis model uses the ResNet50 model architecture implemented in ArcGIS API for Python.Accuracy metricsThis model has an overall accuracy of 97.88 percent. The confusion matrix below summarizes the performance of the model on the validation dataset. Sample resultsHere are a few results from the model:Ground truth: Apple_black_rot / Prediction: Apple_black_rotGround truth: Potato_early_blight / Prediction: Potato_early_bightGround truth: Raspberry_healthy / Prediction: Raspberry_healthyGround truth: Strawberry_leaf_scorch / Prediction: Strawberry_leaf_scorch

A dataset of 61,486 images of plant leaves and backgrounds, with each image labeled with the disease or pest that is present. The dataset was created by researchers at the University of Wisconsin-Madison and is used for research in machine learning and computer vision tasks such as plant disease detection and pest identification.

1) Data Introduction • The Egyptian Plant Leaf Image Dataset (EPLID) is a plant leaf image-based computer vision dataset developed for the purposes of monitoring plant health, detecting diseases, and identifying plant species. It consists of eight distinct classes — Apple, Berry, Fig, Guava, Orange, Palm, Persimmon, and Tomato — each organized into separate folders for convenient labeling and model training.

2) Data Utilization (1) Characteristics of the Egyptian Plant Leaf Image Dataset (EPLID): • The dataset contains images captured in real-world conditions, making it well-suited for the development of AI models that can be applied in practical agricultural environments. • Each image clearly presents the leaf's texture, color, and venation, enabling high-precision applications in plant recognition and disease detection tasks.

(2) Applications of the Egyptian Plant Leaf Image Dataset (EPLID): • Development of AI models for plant disease detection: The dataset can be used to train deep learning models that automatically identify plant diseases by learning abnormal leaf patterns such as spots, discoloration, and surface damage. • Construction of crop classification and cultivar identification systems: The dataset can support the development of models that classify different crop types and identify plant varieties based on their leaf characteristics.

This dataset consists of five subsets with annotated images in COCO format, designed for object detection and tracking plant growth: 1. Cucumber_Train Dataset (for Faster R-CNN) - Includes training, validation, and test images of cucumbers from different angles. - Annotations: Bounding boxes in COCO format for object detection tasks.

1. Tomato Dataset
2. Contains images of tomato plants for 24 hours at hourly intervals from a fixed angle.

3. Annotations: Bounding boxes in COCO format.

4. Pepper Dataset

5. Contains images of pepper plants for 24 hours at hourly intervals from a fixed angle.

6. Annotations: Bounding boxes in COCO format.

7. Cannabis Dataset

8. Contains images of cannabis plants for 24 hours at hourly intervals from a fixed angle.

9. Annotations: Bounding boxes in COCO format.

10. Cucumber Dataset

11. Contains images of cucumber plants for 24 hours at hourly intervals from a fixed angle.

12. Annotations: Bounding boxes in COCO format.

This dataset supports training and evaluation of object detection models across diverse crops.

This is a comprehensive version of the Eggplant Leaf Image Dataset, designed to support machine learning and deep learning research in agriculture, plant pathology, and computer vision. This dataset addresses class imbalance and model generalization challenges by including a significantly expanded collection of images through controlled data augmentation.

The dataset includes a total of 2,180 high-resolution images (6000×4000 pixels), categorized into six disease or health classes of Solanum melongena (eggplant) leaves:

All original images were captured using a Canon EOS 1300D DSLR camera under consistent natural lighting conditions. Files are saved in JPG format, and image resolution is preserved within ±5% of the original dimensions to maintain visual fidelity.

To improve dataset usability for robust model training and generalization, controlled data augmentation was applied using the Albumentations library. The transformations include random rotation, horizontal flipping, brightness/contrast adjustments, slight color shifts, and padding to maintain aspect ratio. All augmentation procedures were consistently applied and seeded for reproducibility. Augmentation parameters are documented in detail in the metadata.

The metadata.csv file provides a class-wise summary including original image count, augmented image count, augmentation ratios, and the exact augmentation pipeline used. The augmentation was seeded for reproducibility.

Note: The original and augmented images are stored in separate folders under the "Original" and "Augmented" directories, respectively. Each directory is organized into six class-specific subfolders: Healthy, Insect-Pest, Leaf-Spot, Mosaic-Virus, Small-Leaf, and Wilt. Augmented images are clearly distinguishable by the inclusion of the substring "_aug_" in their filenames. This clear separation ensures reproducibility, transparency in data provenance, and ease of use for researchers who may wish to train models using only original, only augmented, or both types of data.

Files:

• EggplantLeaf-ImageDataset.zip — Contains all files and folders, inclusind Original, Augmented, metadata and readme.
• OriginalC — Contains only raw field-captured images grouped by class.
• Augmented — Contains synthetically expanded datasets, also organized by class. Augmented filenames include the marker "aug" for easy identification.
• metadata.csv — Class-level summary and augmentation details.
• Readme.md — Technical documentation and usage notes.

The dataset contains 30 types of plants images, including 21000 training images, 3000 validation images and 6000 test images, with a total data size of 1.48GB, and supports the recognition of the following plants types: aloevera, banana, bilimbi, cantaloupe, cassava, coconut, corn, cucumber, curcuma, eggplant, galangal, ginger, guava, kale, longbeans, mango, melon, orange, paddy, papaya, peperchili, pineapple, pomelo, shallot, soybeans, spinach, sweetpotatoes, tobacco, waterapple, and watermelon.

This plant image dataset consists of 14,790 images categorized into 47 distinct plant species classes. The dataset was compiled by collecting images from Bing Images and manually curating them, although not by professional biologist. I collected this images for a project aimed at classifying plant species as either toxic or safe for cats.

Key Features: - Total Images: 14,790 - Number of Classes: 47 - Image Source: Collected from Bing Images - Curation Method: Manual cleaning by non-expert

Dataset Composition: - The number of images per class varies significantly, ranging from 66 (Yucca) to 547 (Monstera Deliciosa). - Some well-represented classes include Chinese evergreen (514 images), Dumb Cane (541 images), and Monstera Deliciosa (547 images). - Classes with fewer images include Yucca (66 images), Kalanchoe (130 images), and Asparagus Fern (169 images).

Image Characteristics: - Images vary in quality and resolution. - The dataset includes both whole plant images and close-ups of specific plant parts. - Plants are placed indoor and outdoors - Images are organized into separate folders for each plant category.

The current dataset is for personal use only due to copyright considerations.

• In our surroundings, there are various types of plants. In our daily lives, we use products derived from these plants in many ways. We depend on a vast kingdom of plants to meet all the needs of life, including food, clothing, shelter, education, and healthcare. We have already gained a lot of knowledge about this relationship. We have seen or heard that when children have a cold or cough at home, they are given a mixture of tulsi (holy basil) leaves and a few drops of honey. As a result, their cold and cough subside, and they feel relieved. If someone suddenly gets a cut on any part of their body, washing the wound with the juice of the aloe vera leaf or applying a poultice of durba grass can be beneficial. This helps stop bleeding, and within two to three days, the wound dries up, and the person recovers. The plants in our environment that are used for the relief or cure of diseases are called medicinal plants. • At one time, Bangladesh was rich in medicinal plants. Fields, riverbanks, roadsides, and forests were abundant with numerous medicinal plants. Due to the increase in population, the diverse use of land has risen. Additionally, due to ignorance, negligence, and lack of attention, the primary habitat of these medicinal plants, natural forest land, has decreased. Consequently, valuable tree resources like these have been diminished. Many species have already become extinct. Despite this, our country still has a sufficient number of medicinal plants scattered across remote and less-explored areas. We are not familiar with all of them. It is crucial for us to be aware of these medicinal plants, recognize them, and understand their uses and properties. As a result, we can play a significant role in the holistic well-being of the general population in our country by contributing to the management and cure of various diseases. • This medicinal plant identification dataset would likely consist of a collection of images and associated metadata related to various medicinal plants. The dataset would serve as a resource for developing and training machine learning models for the automatic identification and classification of medicinal plants. • Six distinct kinds of medicinal plants are shown in this large dataset, which can be used to develop machine vision-based techniques: Arjun Leaf, Curry Leaf, Marsh Pennywort Leaf, Mint Leaf, Neem Leaf, and Rubble Leaf. • In reality, 1380 images of medicinal plants were initially collected from the field. Subsequently, to increase the quantity of data points, various image processing techniques were applied, such as shifting, flipping, zooming, shearing, brightness enhancement, and rotation, resulting in a total of 9660 augmented images derived from the original images.

Plant stress identification dataset comprising RGB and thermal images of five tropical fruit plants Custard apple, Guava, Mango, Lemon, and Sapodilla. Useful for researchers working in the field of plant stress analysis to develop deep learning based models for the plant stress classification task.

Context

This is the dataset that I used in my iOS and Android plant disease detection app, PlantifyDr. You can check out my full open-source project here: https://github.com/lavaman131/PlantifyDr

Content

The dataset contains over 125,000 jpg images of 10 different plant types: Apple, Bell pepper, Cherry, Citrus, Corn, Grape, Peach, Potato, Strawberry, and Tomato. The total number of plant diseases is 37. Augmentations have already been applied to the data, but feel free to add your own augmentations if you like.

Acknowledgements

Special thanks to: https://data.mendeley.com/datasets/tywbtsjrjv/1 https://www.kaggle.com/vipoooool/new-plant-diseases-dataset https://github.com/pratikkayal/PlantDoc-Dataset https://data.mendeley.com/datasets/3f83gxmv57/2

for the data.

Inspiration

The Food and Agriculture Organization of the United Nations (FAO) estimates that annually between 20 to 40 percent of global crop production is lost. Each year, plant diseases cost the global economy around $220 billion. I hoped to use deep learning to solve this problem and be able to better educate farmers and the public with the necessary knowledge to treat their plants.

This dataset is a collection of 13,597 images of cotton leaf surfaces acquired with a hand-held microscope to develop deep learning models to classify images based on leaf hairiness and assist Cotton breeders in their variety selection efforts.

These images were collected from 27 genotypes grown across 2 seasons (2019-2020 and 2020-2021), 2 sites (Australian Cotton Research Institute, -30.21, 149.60, Narrabri, NSW, Australia and CSIRO Black Mountain Laboratories, -35.27, 149.11, Canberra, Australian Capital Territory, Australia) and two growth conditions (Field and Glasshouse). Genotypes have been anonymized to protect germplasm Intellectual Property.

Note 1: This dataset was released with our HairNet paper (Rolland et al 2022, see link below). At the time of publishing Rolland, V., Farazi, M.R., Conaty, W.C. et al. HairNet: a deep learning model to score leaf hairiness, a key phenotype for cotton fibre yield, value and insect resistance. Plant Methods 18, 8 (2022). https://doi.org/10.1186/s13007-021-00820-8, this dataset was called 'Cotton leaf surface image dataset to build deep learning models for leaf hairiness trait (2019-2021)'. It has since being renamed 'CotLeaf-1: Cotton Leaf Surface Images dataset 2019-21'.

Note 2: if you intend to use this dataset in conjunction with CotLeaf-2, CotLeaf-X or AnnCoT datasets, then use the CotLeaf-1 Json file attached to the CotLeaf-2 collection (see link below).

See below for related Datasets and Publications.

Lineage: Genotype Selection: A total of 27 Gossypium hirsutum Cotton genotypes were selected based on their known leaf hairiness. Genotypes were anonymised to protect germplasm intellectual property. Various combinations of these genotypes were grown at two different Australian sites (Narrabri, New South Wales & Canberra, Australian Capital Territory), in the field or controlled glasshouse environment, and over two years (2019-2020 and 2020-2021). For details refer to Rolland et al 2021 (link attached to this submission).

Field experiments - Narrabri Seed of selected genotype were planted on Oct. 21 2019 and Nov. 6 2020, at planting density of 10 - 12 plants m-2 in rows spaced at 1 m. Each genotype was grown in a single 13 m row. The soil of the site is a uniform grey cracking clay. Nitrogen was applied as anhydrous ammonia approximately 12 weeks before planting at a rate of 200 kg N ha-1. Plants were furrow irrigated every 10 to 14 d (approximately 1 ML ha-1 applied at each irrigation) from December through to March, according to crop requirements. Each experiment was managed according to its individual requirements for irrigation and pest control, with all plots receiving the same management regime.

Glasshouse experiments - Narrabri Plants were grown in temperature-controlled glasshouses. About 15 seeds of each genotype were sown in 8 L plastic pots filled with soil on Sept. 6 2019 and Nov. 2 2020, respectively. The soil was obtained from cotton fields as above. To improve the nutrient status of the potting mix 10 g of MULTIgro® basal fertiliser was dissolved into the soil before planting. A 10 mm layer of sand was added to the surface of the pots to reduce surface evaporation and assist in seedling emergence. Once emerged seedlings had reached the three-leaf stage, pots were thinned down to two plants per pot. Plants were grown at 18 °C night and 32 °C during the day, under natural light conditions.

Glasshouse experiment - Canberra Plants were grown in temperature-controlled glasshouses. Eight seeds of selected genotypes were sown in 5 L plastic pots filled with potting mix on Nov. 30 2020. The pots were filled with a 60:40 compost:perlite soil mix. Osmocote® Exact Standard 3-4M was sprinkled on the top layer of soil before flowering. Two weeks after sowing, pots were thinned down to two plants per pot. Plants were grown at 18 °C night and 28 °C during the day, under natural light conditions.

Leaf selection and harvesting: Leaves were numbered in ascending number from the tip of the main stem, with the first fully opened leaf called leaf one. Leaves 3 and 4 from ten individual plants were harvested by cutting their petiole in a proximal position. Harvested leaves were placed in paper bags and imaged within the same day. In the 2019-2020 glasshouse experiment, a few plants died or had a missing leaf, in which case there may be genotypes for which leaves 3 and 4 were harvested from less than 10 plants.

Leaf imaging: Single leaves were imaged at a magnification of about 31x with a portable AM73915 Dino-lite Edge 3.0 microscope equipped with a RK-04F folding manual stage and connected to a digital tablet running DinoCapture 2.0. Images were captured on the abaxial side of the leaf, along the 3 central mid-veins. An average of 3 to 5 images were captured in a proximal to distal fashion along each one of the 3 mid-veins, yielding a total of about 9 to 15 images per leaf. The exact angle of the mid-vein in each image was not fixed. However, either end of the mid-vein was always cut by the left and right borders of the field of view, and never by the top and bottom ones.

Dataset Card for AGM Dataset

  Dataset Summary

The AGM (AGricolaModerna) Dataset is a comprehensive collection of high-resolution RGB images capturing harvest-ready plants in a vertical farm setting. This dataset consists of 972,858 images, each with a resolution of 120x120 pixels, covering 18 different plant crops. In the context of this dataset, a crop refers to a plant species or a mix of plant species.

  Supported Tasks

Image classification: plant phenotyping… See the full description on the dataset page: https://huggingface.co/datasets/deep-plants/AGM.

Invasive Plant Species Detection

## Overview

Invasive Plant Species Detection is a dataset for classification tasks - it contains Species annotations for 1,398 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

In this research, we present a technique aimed at identifying the evolving sections of plants utilizing RGB-D data, with the aim of automating the detection of plant growth within an extraterrestrial experimental setting. As humanity entertains the prospect of inhabiting space in the future, the cultivation of plants in outer space becomes imperative for sustaining food supplies. However, the feasibility of growing plants in space akin to terrestrial methods remains uncertain, necessitating exploration through cultivation experiments conducted aboard international space stations and similar platforms. The observation of plant growth in space is constrained by human resources and available measurement space, further compounded by the exorbitant transportation costs, which escalate with weight. Consequently, there is a preference for lightweight equipment. Traditional automatic plant growth measurement techniques often rely on bulky equipment or require a significant amount of measurement space, rendering them impractical for space applications. In this investigation, we propose a methodology for identifying growing plant sections employing just one RGB-D camera. This approach enables the construction of a measurement system utilizing only a single camera and a laptop for image storage and connection, thereby ensuring lightweight portability. Moreover, the fixed positioning of the camera for plant capture minimizes spatial requirements and reduces the need for manpower. Our proposed technique entails leaf segmentation through depth data and the detection of growing sections via local feature matching. Experimental trials using a model plant corroborated the effectiveness of our method in leaf segmentation and growing part detection. Additionally, the experimental outcomes showcased the capability of the proposed approach in pinpointing the growing sections by refining the matching areas based on segmentation outcomes and appropriate observation intervals.

The Australian National Botanic Gardens (ANBG) has for many years been building up a collection of photographs and illustrations of Australia's native plants. Originally these were photos taken on field collecting trips to supplement the herbarium specimens and data associated with the living plants. The collection attempts to have photos linked to herbarium voucher specimens lodged within one of the major herbaria in Australia. In recent years many other photographers have contributed to the APII with the identification of the plants based on in-situ plant identification in botanic gardens, identification keys, or consulting experts in the area. There over 28,000 Australian plant photos from APII that were either born-digital or have been digitised from slides. There are over 48,000 plant photos held in the APII collection, including those still in slide form and not yet digitised. Searches can be made for all records, with those that are digitised being displayed as thumbnails. Historic photos of the ANBG, other Botanic Gardens, National Parks, landscapes, animals are also held in the collection.

A collection of non-framed multi-spectral images of tomato plants infected with the Tuta Absoluta leafminer disease.

General Information

Instances in the Jacobaea vulgaris class: 895Instances in the Meadow class: 9141Image sizes from 77x77 to 817x817 pixels on three color channels (RGB)

Data Generation and Source

The images in this dataset were taken as part of the project “UAV-basiertes Grünlandmonitoring auf Bestands- und Einzelpflanzenebene” (engl. “UAV-based Grassland Monitoring at Population and Individual Plant Level”), financed by the Authority for Economy, Transport, and Innovation of Hamburg. In September 2018, flights with an octocopter were conducted over two extensively used grassland areas in the urban area of Hamburg. The multicopter flew in a height of circa 11 meters and took pictures with a ground resolution of approximately 3,18 mm/pixel. Additional information about the process of image generation for this dataset are to be found in the relevant papers written by P. Zacharias: 1) UAV-basiertes Grünland-Monitoring und Schadpflanzenkartierung mit offenen Geodaten [p. 45–53] and 2) UAV-basiertes Grünlandmonitoring auf Bestands- und Einzelpflanzenebene.

Additionally, to the images of Jacobaea vulgaris taken by the UAV, the dataset includes images of Jacobaea vulgaris plants from the internet (included in the total 895 images; e.g. images 'jkk0523.jpg', 'jkk0527.jpg'). Furthermore, some of the images of the Jacobaea vulgaris plants have been rotated, further cropped or a filter has been applied. The exact number of augmentations made is unknown. As there are augmented images included in the datasets -which makes the dataset useful for training and validation- a use of the dataset for testing purposes is not recommended due to the risk of data leakage.

Data License

The dataset is licensed under the license CC BY 4.0. The attributor of the data is the Chair of Geodesy and Geoinformatics at the University of Rostock. The data was created within the scope of the project 'UAV-based Grassland Monitoring at Population and Individual Plant Level', financed by the Authority for Economy, Transport, and Innovation of Hamburg.