Description of data collection: 3.1 Experimental site The field experiment was carried out in Pirassununga, São Paulo, Brazil (21º57’31” S, 47º27’07” W, 620 m a.s.l.). In the experimental area, the slope is moderately undulating, and the soil is classified as Rhodic Hapludox. The studied species were Urochloa hybrid Mavuno, common name Mavuno grass, registered in the Ministry of Agriculture, Livestock and Food Supply as MIXE DRWN 12, no 30488. Soil sampling was carried out individually at each experimental unit (plots of 5 x 4 m) in a 0-20 cm soil depth. The results from chemical analysis were used to define the amount of limestone to increase and/or maintain soil base saturation (BS) to 60%. Phosphate and potassium fertilization (using granular simple superphosphate and potassium chloride, respectively) for soil correction were applied individually at each plot aiming to maintain at least 15 mg dm-3 of soil P content and 1.6 mmolc dm-3 of soil K content. Soil correction procedures were implemented from July to October of 2020. Mavuno grass pastures were monitored along successive regrowth cycles from December 2020 to April 2021. The plots were cut always they reached 40 cm, and 20 cm of residual height was maintained in all treatments. Measurements of canopy height were taken weekly after each cutting until the plots have reached the defined pre-cutting target. The height was defined as the average of 10 systematic readings along two transect lines into the plot, using a graduated measuring stick. Treatments were comprised of four N rates, using urea (45% N), applied at the beginning of regrowth and after each cutting procedure following a randomized complete block design with four replications: no-nitrogen (D0), 15 (D15), 30 (D30) and 45 (D45) kg ha-1 of N. The N rates were defined to generate gradients of leaf N contents to fit in three main classes: deficient, moderately deficient, and sufficient. Images aquisition To determine leaf nitrogen (Nleaf, in %), 50 diagnostic leaves (youngest completely expanded leaves) were collected from each plot at the pre-cutting at each regrowth cycle. After being detached from the tillers, the diagnostic leaves were placed side by side on a white matte background clipboard. Each image, containing three leaves, was captured at 23 cm from the base using with a Samsung Galaxy J5 Duos SM-J500M/D smartphone, with a 13 MP camera, stored as 24-bit color images, with a resolution of 800 x 600 pixels and saved in JPEG format. A total of 5 image subsamples with 3 leaves were captured at each regrowth cycle of the plots always between 11 am and 1 pm, with the sun closest to the vertical (Zenith). The smartphone was configured with ISO 100 (camera sensitivity level to available light), lighting 100 and white balance 6500 K. After images acquisition, samples were submitted to drying in a forced-air oven at 55°C, for at least 72 h and, subsequently, ground. Nitrogen content was determined by the method of Kjeldahl. The critical leaf N content which represent the sufficient class for Mavuno grass ranged from 19 to 21 g N kg-1 leaf dry matter (1.9 to 2.1%). Data access and security To ensure the data security, all information regarding the field data collection and images acquired from each plot were saved in the cloud under resticted acess. The data were backed up regularly at both the cloud storage and in an external hard drive to prevent loss. This frequency ensured that changes or additions to the data were captured in the backups. The researcher was the responsible for the backup and recovery process. Regular checks were performed to ensure the integrity of the backups and their ability to restore the data in case of an incident. Anti-virus software were installed on all devices used to access the data to protect against malware and other security threats. No events of data loss or corruption were registered. Description of data set: a. Regrowth cycle: Once the cutting criteria was the sward height of 40 cm, each plot reached the target at different dates. The regrowth cycles were identified as follow: • 0, representing the cut in which the regrowth progressed during the dry season. After this cutting procedure the first N rate was applied to the plots according the respective treatment. • 1, representing the first cutting of the growing season, after the first N rate application. After this cutting procedure the second N rate was applied to the plots according the respective treatment. • 2, representing the second cutting of the growing season, after the second N rate application. After this cutting procedure the third N rate was applied to the plots according the respective treatment. • 3, representing the third cutting of the growing season, after the third N rate application. After this cutting procedure no N was applied to the plots as the dry season had beggining. b. Plot number: The experimental area was comprised by 16 plots of of 5 x 4 m. Each plot was individually numbered from1 to 16. c. Block number: The treatments were assigned to the experimental area according to a completely randomized block desing. The blocking criteria was the initial soil fertility. There were 4 block (replications), numbered from 1 to 4. d. N rate: Representing the field treatments, which were comprised of four N rates : no-nitrogen (0), 15 (15), 30 (30) and 45 (45) kg ha-1 of N. The fertilizer used was urea (45% N), applied at the beginning of regrowth and after each cutting procedure, excepting after the regrowth cycle 3. e. Image number: At each pre cutting stage, the diagnostic leaves were detached from the tillers and imediatelly placed side by side on a white matte background clipboard. Each image, containing three leaves, was captured at 23 cm from the clipboard using a smartphone. A total of 5 image subsamples with 3 leaves were captured, and images were numbered from 1 to 5. In the cases in which images were considered not adequate, they were identified using the code “NA”. f. Image number: At each pre cutting stage, the diagnostic leaves were detached from the tillers and imediatelly placed side by side on a white matte background clipboard. Each image, containing three leaves, was captured at 23 cm from the clipboard using a smartphone. A total of 5 image subsamples with 3 leaves were captured, and images were numbered from 1 to 5. In the cases in which images were considered not adequate, they were identified using the code “NA”. g. Sampling dates: Once the cutting criteria was the sward height of 40 cm, each plot reached the target at different dates. Thus, sampling dates for each plot were registered as day/month/year. h. Nitrogen concentration: The N concentration was determined in a composed sample, representing the 50 diagnostic leaves (youngest completely expanded leaves) collected from each plot at the pre-cutting at each regrowth cycle. Nitrogen content was determined by the method of Kjeldahl. Values are expressed in %. i. Class: Bsed on the results of the laboratorial analysis of N concentration, samples were classified into defficient (D, when N content was lower than 1.9%) and sufficient (S, when N content was higher than 1.9%).