This paper focuses on predicting colour trends in fashion, a constantly evolving form of human expression. Current literature lacks methods to effectively detect and analyze colours in fashion apparel images for accurate trend forecasting. To address this gap, the study collects apparel images from different regions of India to extract and analyze the dominant colours specific to India. The proposed approach uses YOLO object detection to identify human key points within the images, followed by isolating the top and bottom wear. These areas are analyzed using the Colour Thief algorithm to extract dominant colours, which are then converted into hex codes and mapped to corresponding colour names. These colours are further grouped into broader categories and analyze as time series. Using Patch transformers (PatchTST), the model predicts future colour trends for the upcoming seasons. This system aims to assist Indian retailers by providing insights into emerging colour patterns, enabling them to better plan and strategize their merchandise offerings.