Deep Learning

"Multiclass classification of damages caused by natural disasters". Here I used DeepLabV3+ for semantic segmentation & Custom CNN for classification. I want to make some changes & updation so that I will increase the model performance like accuracy & F1-Score because the dataset is imbalanced. I want F1-score & accuracy around 85-90%. Code is there means more than 70% of the work is already done , just hypertunning and image manipulation work is remaining.My contact no. 9922700207. My budget 3k-5k.