This study aims to enhance the performance of organic to recyclable waste classification using deep learning techniques. The negative impacts of poor waste segregation schemes on environmental and social development are highlighted, emphasizing the need for an automated segregation process. Manual waste classification is time-consuming, costly, and less accurate. The proposed work employs an Improved Deep Convolutional Neural Network (DCNN) to automate the segregation process. The dataset consists of 25,077 images divided into 70% training and 30% testing images. The performance metrics used are classification accuracy, Missed Detection Rate (MDR), and False Detection Rate (FDR). The results show that the proposed model achieves a classification accuracy of 93.28%, outperforming other models such as VGG16, VGG19, MobileNetV2, DenseNet121, and EfficientNetB0. The study also compares the impact of different hyper-parameters and regularization techniques, demonstrating the effectiveness of the proposed model in reducing MDR and FDR while maintaining high accuracy. The research contributes to the field of waste management by providing a more efficient and accurate solution for waste classification and recycling.This study aims to enhance the performance of organic to recyclable waste classification using deep learning techniques. The negative impacts of poor waste segregation schemes on environmental and social development are highlighted, emphasizing the need for an automated segregation process. Manual waste classification is time-consuming, costly, and less accurate. The proposed work employs an Improved Deep Convolutional Neural Network (DCNN) to automate the segregation process. The dataset consists of 25,077 images divided into 70% training and 30% testing images. The performance metrics used are classification accuracy, Missed Detection Rate (MDR), and False Detection Rate (FDR). The results show that the proposed model achieves a classification accuracy of 93.28%, outperforming other models such as VGG16, VGG19, MobileNetV2, DenseNet121, and EfficientNetB0. The study also compares the impact of different hyper-parameters and regularization techniques, demonstrating the effectiveness of the proposed model in reducing MDR and FDR while maintaining high accuracy. The research contributes to the field of waste management by providing a more efficient and accurate solution for waste classification and recycling.