Abstract: Lighting is one of the important environmental factors in pig breeding. This study aims to clarify the optimal light intensity for gilts in an artificial light environment. A test house was constructed to evaluate the preference of gilts for light intensity. LED white light (400-700 nm) was used as the artificial light source in the experimental pigsty. Four light intensities were adopted as 40, 100, 350, and 1200 lx in the test. According to the light intensity, 40 and 100 lx were defined as the group of weak light intensity, while 350 and 1200 lx were the group of strong light intensity. The light scheme was adjusted for each pen every seven days in a cycle, with a light duration of 16 h (lights on at 05:00 and off at 21:00). A total of 24 gilts with similar ages and weighs were selected from the same pigsty as the experimental subjects. A 4-d pre-experiment and a 28-d trial were then conducted to explore the preference selection of gilts for the light intensity in an artificial light environment. The evaluation indicators were taken as the spatiotemporal distribution, behavioral activity level, feeding amount, and feeding frequency of the gilts. The industrial cameras and YOLOv8n structure were used to record and analyze the activity status of pigs during the experiment. The results showed that the average accuracies of the trained image recognition model for pig standing, sitting, lying and feeding behavior recognition were 97.5%, 90.3%, 98.5% and 95.2%, respectively. According to the spatiotemporal distribution of the pig population, there was a significant preference for a light intensity of 100 lx in the lighting period (P<0.05). The number of pigs also accounted for 29.33% ± 1.14% of the total. Furthermore, the rhythmic patterns were found in the preference for the rest light intensities. There was a significantly higher preference for 40 lx in the morning and evening periods, compared with 350 lx and 1200 lx (P<0.05), corresponding to the week light environment of the transition between morning and evening in nature. The activity levels were also determined for six hours, after the lights were turned on. The pigs shared significantly higher activity levels in the higher light intensity groups (350 lx: 37.97% ± 3.47% and 1200 lx: 35.42% ± 4.04%), compared with the lower ones (40 lx: 27.90% ± 8.44% and 100 lx: 23.94% ± 3.79%). Since most of the pigs were lying in the house, the overall distribution of pigs in the inactive state was similar to the number distribution of pigs. While there was no outstanding number of active pigs in each light intensity. The pigs also preferred to take a break at noon. In free feeding, the feed consumption was the highest under 1200 lx illumination, followed by 350 lx group. In feeding behavior, the average daily feeding time of each gilt was (60.64±1.31) min in this experiment. And the feeding time of gilts under the 350 lx and 1200 lx was significantly longer than that of 40 lx (P<0.05). Therefore, it was recommended to gradually adjust the lighting intensity using a dim-to-bright approach, with a primary light intensity of 100 lx. During the 3 hours before and after the lights were turned on, a lower light environment of 40 lx was used to avoid excessive light stimulation. Additionally, the 2 h duration of higher light intensity (350 lx) can promote the pig activity levels. The finding can also provide a theoretical foundation for the precise control of the lighting environment for gilt facilities.