The massive volume and multiple dimensions of germplasm data have caused the low efficiency of data mining. This problem is envisaged in this paper by presenting a Hadoop-based data mining platform for crops germplasm after detailed analysis of both the components and working mechanism of Hadoop cloud platform. Different functional modules will also be discussed in detail, based on which, specific development programs of the platform are also to be proposed. The efficiency and feasibility of the platform will be verified through efficiency tests of improved classic Apriori algorithm.