The paper provides a comprehensive review of the evolution and advancements in object detection over the past two decades, from the 1990s to 2022. It highlights the significant impact of deep learning on the field, transitioning from traditional handcrafted features to deep convolutional neural networks (CNNs). Key milestones include the Viola-Jones detector, HOG detector, deformable part-based models (DPM), and the introduction of CNN-based detectors like RCNN, SPPNet, Fast RCNN, and Faster RCNN. The paper also discusses the evolution of multi-scale detection, context priming, hard negative mining, loss functions, and non-maximum suppression (NMS). Additionally, it covers speed-up techniques such as feature map sharing, cascaded detection, network pruning, lightweight network design, and numerical acceleration. The recent advances in object detection are reviewed, focusing on state-of-the-art methods like YOLO, SSD, RetinaNet, CornerNet, CenterNet, and DETR, which have achieved significant improvements in accuracy and speed. The paper concludes with a discussion on future research directions and potential applications of object detection in various fields.The paper provides a comprehensive review of the evolution and advancements in object detection over the past two decades, from the 1990s to 2022. It highlights the significant impact of deep learning on the field, transitioning from traditional handcrafted features to deep convolutional neural networks (CNNs). Key milestones include the Viola-Jones detector, HOG detector, deformable part-based models (DPM), and the introduction of CNN-based detectors like RCNN, SPPNet, Fast RCNN, and Faster RCNN. The paper also discusses the evolution of multi-scale detection, context priming, hard negative mining, loss functions, and non-maximum suppression (NMS). Additionally, it covers speed-up techniques such as feature map sharing, cascaded detection, network pruning, lightweight network design, and numerical acceleration. The recent advances in object detection are reviewed, focusing on state-of-the-art methods like YOLO, SSD, RetinaNet, CornerNet, CenterNet, and DETR, which have achieved significant improvements in accuracy and speed. The paper concludes with a discussion on future research directions and potential applications of object detection in various fields.