The technique of motif extraction comprises replicating painted rock art without the rock's natural background texture. Both manual and computerized methods are acceptable. Today's most popular way of extracting motifs from rock art is employing color selector tools in programs like Adobe Photoshop after applying post-processing techniques to the photos such as D-Stretch software. Before motif analysis, which is the fundamental part of a comprehensive understanding of rock art panels, motif extraction is an essential step. In the Central Indian region of Madhya Pradesh, painted rock shelters are present in large numbers. With hundreds of shelters and thousands of photographs collected, it is laborious, repetitious, and time-consuming to manually or digitally (with manual color selection) extract themes from the images for analysis. Due to the lack of time and resources, it is imperative to develop an automated technique for extracting motifs from photographs of rock art to assist researchers in investigating painted rock surfaces. As part of our research, we compiled a training dataset that included input images and desired outputs from select rock shelters in Narmadapuram and Sehore districts. Once the model was familiar with all the data, we tested it by feeding it an image outside the training dataset. Here, we try to create a proof-of-concept machine learning (ML) model and train it to recognize motifs in photos of rock art and extract them. Using information gathered from painted open-air rock shelters in the Narmada Basin region of Central India, this model is trained, and its effectiveness is assessed and presented here. Once it has been adequately trained in the motif patterns found in those specific locations, the same model may be used and applied to extract motifs from rock art sites all over the world.