In order to actively promote the construction of low-carbon universities, taking the Yanyuan Campus of Peking University as a case study, we utilized the carbon emission factor method to calculate the total campus carbon emissions for 2023. The results are as follows. Firstly, total carbon emissions of Peking University main campus for 2023 exceeded 250000 tons, with Scope 1 (direct emissions) accounting for 17.20%, Scope 2 (indirect emissions from purchased energy) for 55.31%, and Scope 3 (other indirect emissions) for 27.49%. Secondly, compared with the carbon emissions in 2009, Scope 1 and Scope 2 carbon emissions increased by 16.76%, while per capita carbon emissions decreased by 19.10%. Based on the calculated carbon emission scenarios, we propose carbon reduction recommendations targeting the three major sources of carbon emissions at Peking University main campus — energy consumption, transportation, and food consumption. It is estimated that the potential to reduce carbon emissions in energy consumption is more than 30%, more than 5% in transportation, and more than 3% in food consumption. The total carbon reduction potential in these three areas is close to 40%, equivalent to over 120000 tons of carbon emissions annually. 

At present, building extraction from high-resolution remote sensing images using deep neural network is viewed as a binary classification problem, which divides the pixels into two categories, building and nonbuilding, but it cannot distinguish individual buildings. To solve this problem, the U-Net modified with Xception module and multitask learning are combined to apply to the instance segmentation of buildings, which both acquires the binary classification and distinguishes the individual buildings. Inria aerial imagery is used as the research dataset to validate the algorithm. The results show that the binary classification performance of U-Net modified with Xception outperforms U-Net by about 1.4%. The multitask driven deep neural network not only accomplishes the instance segmentation of buildings, but also improves the accuracy by about 0.5%.