Close　contact　routes,　including　short-range　airborne　and　large-droplet　routes,　play　an　important　role　in　the　transmission　of　SARS-CoV-2　in　indoor　environments.　However,　the　exposure　risk　of　such　routes　is　difficult　to　quantify　due　to　the　lack　of　data　on　the　close　contact　behavior　of　individuals.　In　this　study,　a　digital　wearable　device,　based　on　semi-supervised　learning,　was　developed　to　automatically　record　human　close　contact　behavior.　We　collected　337,056　s　of　indoor　close　contact　of　school　and　university　students　from　194.5　h　of　depth　video　recordings　in　10　types　of　indoor　environments.　The　correlation　between　aerosol　exposure　and　close　contact　behaviors　was　then　evaluated.　Individuals　in　restaurants　had　the　highest　close　contact　ratio　(64%),　as　well　as　the　highest　probability　of　face-to-face　pattern　(78%)　during　close　contact.　Accordingly,　university　students　showed　greater　exposure　potential　in　dormitories　than　school　students　in　homes,　however,　a　lower　exposure　was　observed　in　classrooms　and　postgraduate　student　offices　in　comparison　with　school　students　in　classrooms.　In　addition,　restaurants　had　the　highest　aerosol　exposure　volume　for　both　short-range　inhalation　and　direct　deposition　on　the　facial　mucosa.　Thus,　the　classroom　was　established　as　the　primary　indoor　environment　where　school　students　are　exposed　to　aerosols.