Trash classification is an effective measure to protect the ecological environment and improve resource utilization. With the advancement of deep learning, it is now possible to classify trash using deep convolutional neural network. This work suggests a hybrid deep learning model based on deep transfer learning, which incorporates upper and lower streams, to categorize the trash of the Trash Net dataset, which consists of six classes of rubbish photos. First, the upper stream classifies the input garbage image into either category CGT or category MPP (metal, paper, and plastic class) (cardboard, glass, and trash class). The bottom stream then makes a precise rubbish classification prediction based on the upper stream's findings. In comparison to other state-of-the-art methods, the suggested hybrid deep learning model obtains the best results with a 98.5% accuracy rate. The suggested model may legitimately use the attributes of the image for classification through the verification of CAM (class activation map), which explains why this mode performs better.