The POLLEN73S is an annotated image dataset for the Brazilian Savannah pollen types, which consists of 2523 pollen grain images taken at different angles and distributed in 73 pollen types. The pollen grain examples were obtained from samples flowering species collected in the Brazilian Savannah, more precisely within a radius of 1.5 km from 20°23'16.8"S 54°36'36.3"W coordinates in Campo Grande City urban area, capital of Mato Grosso do Sul State in Brazil.

The POLLEN73S is an annotated image dataset for the Brazilian Savannah pollen types, which consists of 2523 pollen grain images taken at different angles and distributed in 73 pollen types. The pollen grain examples were obtained from samples flowering species collected in the Brazilian Savannah, more precisely within a radius of 1.5 km from 20°23'16.8"S 54°36'36.3"W coordinates in Campo Grande City urban area, capital of Mato Grosso do Sul State in Brazil.

The DBrasiliensis dataset is a repository with 246 images containing 5141 cysts of Dendrocephalus brasiliensis, a native species from South America. The cyst examples were obtained from substrates portions took from the bottom of an aquarium used as an incubator and split into two parts: one to capture the training images and the other to capture the test images.

Training images: the substrate was fixed on white coverslips and observed under a XTRAD USB digital microscope model XT-2036, in which the images were captured using 52x objective lenses. Subsequently, the LabelImg software was used to label cysts in both PASCAL VOC and YOLO formats. There are 111 images in all for training with a resolution of 640x480, containing 3173 annotated cysts.

Test images: the testing set was built with the aim of assigning a weight to a set of images. The substrate was split into small portions on a white coverslip and weigh it using a precision scale. Then, we capture an image of each portion of the substrate using the digital microscope. The captured images were stored in a folder, whose name indicates the amount of cyst and the weight of the substrate contained in the set of images. Besides, the file name of each image in the folder indicates the image number and amount of cyst contained in it. The complete testing set has 1968 cysts in 135 images with a resolution of 640x480 distributed in 10 folders. The substrate weight used to build the testing set is 4.24 grams.

The DBrasiliensis dataset is a repository with 246 images containing 5141 cysts of Dendrocephalus brasiliensis, a native species from South America. The cyst examples were obtained from substrates portions took from the bottom of an aquarium used as an incubator and split into two parts: one to capture the training images and the other to capture the test images.

Training images: the substrate was fixed on white coverslips and observed under a XTRAD USB digital microscope model XT-2036, in which the images were captured using 52x objective lenses. Subsequently, the LabelImg software was used to label cysts in both PASCAL VOC and YOLO formats. There are 111 images in all for training with a resolution of 640x480, containing 3173 annotated cysts.

Test images: the testing set was built with the aim of assigning a weight to a set of images. The substrate was split into small portions on a white coverslip and weigh it using a precision scale. Then, we capture an image of each portion of the substrate using the digital microscope. The captured images were stored in a folder, whose name indicates the amount of cyst and the weight of the substrate contained in the set of images. Besides, the file name of each image in the folder indicates the image number and amount of cyst contained in it. The complete testing set has 1968 cysts in 135 images with a resolution of 640x480 distributed in 10 folders. The substrate weight used to build the testing set is 4.24 grams.