• Dr.Dheeraj Bhimrao Lokhande
• Dr. T.A.Chavan
• Prof.D.P.Patil

One of the most significant uses of deep learning technology is object recognition since it can learn features and capture them in a unique way from more conventional methods. Finding minor size differences in things is the key obstacle in object recognition. A set of multi-scale depth picture features that were taken from the backbone can be used in conjunction with top-down sampled feature pyramid networks to achieve this. Recognition systems use feature pyramids to find objects of various sizes. Pyramid representations have recently been shunned by object detectors that employ deep learning, in part because of how much memory and processing power they tend to demand. In our study, we investigate the potential of multiscale, pyramidal hierarchies to reduce the marginal costs of deep convolution networks. Top-down architecture with lateral connections is used to produce semantic feature maps with detailed information at all scales. With the feature pyramid network acting as a general feature extractor, performance gains are seen. The system is experimentally evaluated as part of the study, and the augmented pyramid network technique outperforms the FCOS model in terms of average precision (AP) by 1.2 on the MS-COCO test-dev. The study's findings showed that a feature pyramid network increased object localisation accuracy.

FPN; Deep learning; Average Precision; Object detection; video detection;

"USE OF A FEATURE PYRAMID NETWORK FOR OBJECT DETECTION- A DEEP LEARNING TECHNOLOGY ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 12, page no.d101-d108, December-2022, Available :http://www.jetir.org/papers/JETIR2212317.pdf

"USE OF A FEATURE PYRAMID NETWORK FOR OBJECT DETECTION- A DEEP LEARNING TECHNOLOGY ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 12, page no. ppd101-d108, December-2022, Available at : http://www.jetir.org/papers/JETIR2212317.pdf