“…However, currently available datasets exhibit multi-scale properties [ [5] , [6] , [7] ], are characterized by classes imbalance [ [5] , [6] , [7] , [8] , 10 ], include few bee species [ [5] , [6] , [7] , [10] , [11] , [12] , [13] , [14] , [15] , [16] , [17] , [18] , [19] , [20] , [21] ], contain a small number of bee samples [ [5] , [6] , [7] , [8] , [9] , 13 , 16 , 18 , 20 , 21 ], lack sufficient standardized and labeled data [ [5] , [6] , [7] , 13 , 15 , 16 , 18 , 20 , 21 ], and have restricted accessibility [ [5] , [6] , [7] , [12] , [13] , [14] , [16] , [17] , [18] , [19] , [20] ], agreeing with [ 3 , 4 ]. The dataset presented in this work overcomes most of these challenges except for including a few bee species, which we aim to address in future work with other case studies and supports our previous study about multiple objects tracking in native beehives [ 1 ].…”