Cotton is one of the most widely grown and commercially significant crop globally, which is majorly utilized in textile production. The Cotton boll detection and counting technique is developed based on Machine learning (ML) and Deep Learning (DL) algorithms. There are limited techniques available for detection of cotton boll count, in which most of the techniques poses several difficulties, such as low performance, high time consumption, and computational complexity and so on.To overcome these kinds of issues, the proposed methodology is developed to provide efficient detection technique for cotton boll count. Initially, the data is gathered from dataset and augment the dataset using Geometric distortion and Photometric distortion approaches. Then, Rolling Joint Bilateral Filter (RJBF) is utilized to pre-process the image to enhance the quality of image. Finally, the Attention based Shuffle Bi-directional feature pyramid network enclosed YOLOv8 (ASB-FPNY)model is utilized to extract the feature from a pre-processed image for cotton boll detection and counting. The progressive attention function is utilized to enhance the performance of ASB-FPNY model for cotton boll detection.The accuracy of proposed technique is 98.18% which is higher than other related techniques. The precision achieved is 98.19%,recall is 97.18%,specificity is 98.2%, and F1-score is 97.68% with the proposed model.