Although much research has focused on CO2 emissions driven by household consumption, significant challenges remain in capturing complex regional variations and the indirect contributions of disaggregated energy sectors and different income groups. In this study, an affinity-propagation multi-regional input-output (AP-MRIO) model is developed through incorporating Dagum Gini coefficient (DGC) and affinity propagation (AP) clustering within a multi-regional input-output (MRIO) modeling framework. AP-MRIO not only traces CO2 emissions from China's provincial household consumption, particularly within disaggregated energy sectors, but also reveals the interaction between sector emissions and income levels. Results obtained disclose that (i) within energy sectors, thermal power, electric power distribution, and petroleum are major emitters (accounting for 68.7 %, 17.5 %, and 6.6 %, respectively); in comparison, CO2 emissions from renewable energy sectors (hydropower, nuclear, wind, and solar) are lower (2.7 %); (ii) urban middle- and high-income households contribute significantly to CO2 emissions (57.1 %), and notable carbon inequalities exist both within and between regions for energy sectors; (iii) some provinces (e.g., Inner Mongolia, Liaoning, and Heilongjiang) should prioritize reducing per capita emissions from the non-renewable energy sectors, and other provinces (e.g., Ningxia, Guangdong, and Hunan) should further promote the development of renewable energy and focus on emissions embodied in the use of intermediate products.