Community　detection　has　been　an　issue　in　complex　network　research.　In　the　paper,　according　to　the　definition　of　weak　community,　we　firstly　propose　a　local　modularity　and　then　design　a　new　mutation　operator　with　better　efficiency　based　on　local　modularity.　The　mutation　operator　selects　the　neighbor　node　that　can　best　embody　the　definition　of　weak　community　structures　as　mutated　result,　which　makes　the　mutated　candidate　solution　closer　to　the　optimal　solution.　Furthermore,　to　accelerate　the　emergence　of　the　optimal　solution,　the　roulette　selection　is　integrated　into　a　uniform　crossover　operator.　On　the　basis　of　the　above　works,　an　improved　Genetic　Algorithm　based　on　the　local　modularity　(IGALM)　is　presented　for　Community　detection.　The　proposed　algorithm　is　tested　and　compared　to　the　other　algorithms　on　both　computer-generated　network　and　real-world　networks.　The　comparative　experimental　results　reflect　that　the　new　algorithm　is　feasible　and　effective　in　small　and　large　scale　complex　networks.