Algorithm 1: Infected Region Detection

INPUT:

TrS, BBx

OUTPUT:

Identified Area, EfDet, Class

TrS—total images used for model training.

BBx—coordinates of the rectangular box showing the diseased portion.

Identified Area—diseased portion in the output.

EfDet –EfficientDet model with the EfficientNet-B0 base network.

Class portion—Label indicating the category of each identified area.

Size_of_Sample ← [x y]

// Computing region of interest

α← AnchorsComputation(TrS, BBx)

// EfDet-Approach

EfDet ← EfficientDet_B0Base (Size_of_Sample, α)

[Ptr Pte] ← Distribution of the employed repository in the train and test parts

// Training phase

For each image i in → Ptr

Compute EffNet(B0)-features→dfCompute EffNet(B0)-features→df

Accomplish keypoints Fusion (df) →Ff

End

Train EffDet using Ff, and measure execution time t_EffDet

η_EffDet← LocateAffectedRegion(Ff)

Ap_EffDet ← Evaluate_AP (EffNet(B0), η_ EffDet)

For each sample j in → Pte

a) Extract key features through trained model Ap_EffDet →βI

b) [Bbox, ConfidenceScore, category] ←Estimate (βI)

c) show the image with Bbox, score, and class label

d) compute

Accuracy mAP precision recall time

End For