Fine-grained Recognition of Plants from Images
Result description
The work focuses on automatic and semi-automatic fine-grained recognition of plants from an image, set of images or videos. Computer vision based plant recognition is a challenging problem due to the variable appearance of plants, high intra-class variability and small inter-class differences, complex geometry and multi-scale hierarchical structure. Plant recognition tasks, ranging from identification of plants from specific plant organs, to general plant recognition and localization "in the wild", are discussed. State-of-the-art methods are reviewed and examples of applications are presented, such as mobile apps for plant recognition, automatic mapping of plant species distribution or automation in agriculture. Our contributions so far include the state-of-the-art methods for leaf and bark recognition using the Fast Features Invariant to Rotation and Scale of Texture, studies of texture- and color-based recognition methods, as well as a deep learning contribution to the PlantCLEF 2016 plant identification challenge, achieving very competitive results. The goals of the thesis are to further study the plant recognition tasks and propose methods suitable for detection, localization and recognition of plants from images at relatively low computational cost, making them suitable for the proposed applications. Extensions to related problems are discussed, including recognition of other natural objects and species, and other fine-grained recognition problems such as food product classification.
Keywords
computer visionrecognitionfine-grainedidentificationplantplantstreesleavesbarkconvolutional neural networksCNNtexture
The result's identifiers
Result code in IS VaVaI
Result on the web
http://cmp.felk.cvut.cz/pub/cmp/articles/sulc/Sulc-TR-2016-09.pdf
DOI - Digital Object Identifier
—
Alternative languages
Result language
angličtina
Original language name
Fine-grained Recognition of Plants from Images
Original language description
The work focuses on automatic and semi-automatic fine-grained recognition of plants from an image, set of images or videos. Computer vision based plant recognition is a challenging problem due to the variable appearance of plants, high intra-class variability and small inter-class differences, complex geometry and multi-scale hierarchical structure. Plant recognition tasks, ranging from identification of plants from specific plant organs, to general plant recognition and localization "in the wild", are discussed. State-of-the-art methods are reviewed and examples of applications are presented, such as mobile apps for plant recognition, automatic mapping of plant species distribution or automation in agriculture. Our contributions so far include the state-of-the-art methods for leaf and bark recognition using the Fast Features Invariant to Rotation and Scale of Texture, studies of texture- and color-based recognition methods, as well as a deep learning contribution to the PlantCLEF 2016 plant identification challenge, achieving very competitive results. The goals of the thesis are to further study the plant recognition tasks and propose methods suitable for detection, localization and recognition of plants from images at relatively low computational cost, making them suitable for the proposed applications. Extensions to related problems are discussed, including recognition of other natural objects and species, and other fine-grained recognition problems such as food product classification.
Czech name
—
Czech description
—
Classification
Type
O - Miscellaneous
CEP classification
JD - Use of computers, robotics and its application
OECD FORD branch
—
Result continuities
Project
GBP103/12/G084: Center for Large Scale Multi-modal Data Interpretation
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
Others
Publication year
2016
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Basic information
Result type
O - Miscellaneous
CEP
JD - Use of computers, robotics and its application
Year of implementation
2016