Grapevines Leaves from Southern Perú -Ds

Description

This study presents a unique and original dataset, composed of 4231 images of grape plants, collected in the Villacuri Valley, Ica, Peru. The images were captured using two different cameras, a Cano and a GoPro, to ensure the diversity and quality of the images. The data collection process was meticulous and well-planned, with the goal of creating a data set that can be reused in various grapevine-related studies. This data set has the potential to be a valuable tool in viticulture research, especially in the study and detection of diseases and pests in grapes, such as Spodoptera and Esca. The data set not only provides a detailed insight into the current conditions of grape plants in the Villacuri Valley, but also opens up new possibilities for the application of modern technologies in viticulture. The potential reuse of this dataset is broad, as it can be used to train machine learning models, perform agricultural data analysis, improve precision farming techniques, and much more.

Steps to reproduce

The dataset presented in this study was developed through a meticulous and well-planned process, with the aim of ensuring its usefulness and applicability in various viticulture-related research. The following describes in detail how the data were collected, the methods and protocols employed, as well as the instruments and workflows used. Location and planning of data collection Data collection was carried out in the Villacurí Valley, Ica, Peru, a region known for its ideal climatic conditions for grape cultivation. Representative vineyard plots were selected, taking into account factors such as soil variability, agricultural management, and the prevalence of specific diseases or pests. Data collection was carried out during different stages of the grapevine growth cycle, ensuring a complete representation of the phenological and phytosanitary conditions of the plants. Image capture methods and protocols To ensure the quality and diversity of the dataset, two different cameras were used: • Canon EOS 5D Mark IV: Selected for its ability to capture high-resolution images and precise plant details, especially under controlled light conditions. • GoPro HERO 9: Used for its versatility in dynamic captures and its ability to operate in field conditions, such as hot climates and uneven terrain. Tagging and classification protocol Each image was manually tagged with relevant metadata, such as: • Exact geographic location using GPS. • Date and time of capture. • Phenological state of the plant. • Presence of diseases (e.g., symptoms of Spodoptera and Esca) or pests. To standardize the tagging, a protocol developed in collaboration with expert grapevine agronomists was followed, guaranteeing the consistency and relevance of the data collected. Software and workflow The workflow for managing and processing the dataset included the following stages: 1. Image management: Photographs were stored and organized using Roboflow software, allowing for minimal adjustments (e.g., exposure and white balance) and ensuring the integrity of the original data. 2. Annotation and labeling: LabelImg was used to make precise annotations on the images. This facilitated the delimitation of areas affected by diseases or pests. 3. Storage and backup: The data was hosted in a cloud storage system (OneDrive), with periodic backups to prevent loss. 4. Data validation: Roboflow was used to verify the quality and consistency of the labels, detecting and correcting possible errors. Reuse potential The design of the dataset follows standards that facilitate its use in future studies. The images are organized in a way that they can be easily integrated into machine learning models, agricultural analysis, or precision agriculture research. Furthermore, access to detailed metadata allows for deeper, more specific exploration based on the needs of each research.

Categories

Funding

Licence