Table 5 Part 2: an analytical study on examples of crop diseases protection and weeds detection approaches; highlights the applied algorithm, plant and data type, data pre-processing and the extracted features
References | Application | Algorithm | Plant | Data type | Data pre-processing | Extracted features |
|---|---|---|---|---|---|---|
[129] | Leaf disease detection | Deep learning CNN: CaffeNet | 13 plants leaves | Images | augmentation by: afine transformation and perspective transformation and rotation manually pre-processing by image cropping and labelling | Â |
Weeds classification | ANN:PSO and bee for optimisation | Potatoes rice | Stereo video | Segmentation | Color features & vegetation indices | |
[128] | Mid-late season weed detection | CNN | Soybean | Aerial images | Overlapped images removal dimenssion reduction annotation | Patches |
[118] | Weed detection | DNN | Sugar beet and weeds | Multi-spectral UAV | Segmentation | RGB Color-Infrared NDVI |
[106] | Leaf disease detection | RF, SVM, KNN | Alfalfa | Digital images | Lesion: artificial cuttingsegmentation:12 lesion segmentation with K-median clustering and linear discriminant analysis | 129 texture colour and shape |
[55] | Seeds disease detection | ANN | Orchids | Digital images | Segmentation: an exponential transform with an adjustable parameter | Texture and colours |
[109] | Plant disease detection | RF | Soybean | Satellite images Crop rotation | Geometric distortions removal radiometrically and sensor correction image rotation | Spectral bands of:red, green, blueNDVI, NIR |
[28] | Leaves disease detection | Transfer learning CNN: abstraction level fusion | Olive | Digital images | segmentation: automatic cropping: Otsu’s algorithm | Edge magnitudes: Gray-scaledShape features: area, perimeter |
[6] | 10 leaves disease detection | Transfer learning CNN | Eggplant, hyacinth beans ladies finger, lime | Digital images | Segmentation data augmentation | / |
[79] | Leaves disease detection | Deep learning: Alex NetGoogLeNet | Apple | Digital images | No pre-processing AlexNet Precursor for features maps max-pooling for GoogLeNet for features extractiondata augmentaion:light disturbance &rotationnoise removal | / |
[81] | Plant disease detection | KNN | Wheat | Satellite imagesfield survey | Radiometric calibration atmospheric correction | Red and green bands NIR, vegetation indices:disease water stress index optimised soil adjusted vegetation index shortwave infrared water stress index triangular vegetation index and others |
[156] | Plant disease detection | RF | Wheat | Satellite images field canopy hyperspectral | Noise removal image mosaicking Atmospheric correction spatial resolution re-sampling | Disease indexNDVi, EVI and others |