With the increasing energy use and carbon emissions in the transportation industry, its impact on the greenhouse effect is gradually being recognized. Therefore, this study aims to explore the achievement of carbon emission peak and carbon neutrality in transportation through prediction. The research employs a deep learning model, the CNN-GRU-Attention model, to predict carbon emissions in the transportation industry in Jiangsu, China. We select influencing factors through an extended STIRPAT model coupled with Lasso regression, and construct the CNN-GRU-Attention traffic carbon emission prediction model according to data indicators from 1995 to 2021. The model predicts carbon emissions from the transportation industry in Jiangsu Province between 2022 and 2035 under six distinct scenarios and proposes corresponding emission reduction strategies. The results show that the model in this study has higher prediction accuracy compared with other models, with a mean absolute error (MAE) of 0.061582, root mean square error (RMSE) of 0.085025, and R² of 0.91609 on the test set. Scenario-based predictions reveal that emission peak in the transportation industry in Jiangsu Province can be achieved under the clean development and comprehensive low-carbon scenarios, with technological innovation being the primary driver of low-carbon emission reductions. This study provides a novel approach for forecasting carbon emissions from the transportation industry and explores the implementation path of emission peak through this method.