# Light-Weight RetinaNet for Object Detection

arXiv: https://arxiv.org/abs/1905.10011 · HTML: https://mlsanity.com/paper/1905.10011/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 24 May 2019.

## Built on by (most cited first, 11 of 11)

- [Top-Down Framework for Weakly-supervised Grounded Image Captioning](https://mlsanity.com/paper/2306.07490/) (2023)
- [WeedVision: Multi-Stage Growth and Classification of Weeds using DETR and RetinaNet for Precision Agriculture](https://mlsanity.com/paper/2502.14890/) (2025)
- [DeepRA: Predicting Joint Damage From Radiographs Using CNN with Attention](https://mlsanity.com/paper/2102.06982/) (2021)
- [You Better Look Twice: a new perspective for designing accurate detectors with reduced computations](https://mlsanity.com/paper/2107.10050/) (2021)
- [Policy-Driven Transfer Learning in Resource-Limited Animal Monitoring](https://mlsanity.com/paper/2509.10995/) (2025)
- [NoFADE: Analyzing Diminishing Returns on CO2 Investment](https://mlsanity.com/paper/2111.14059/) (2021)
- [Knowledge Detection by Relevant Question and Image Attributes in Visual Question Answering](https://mlsanity.com/paper/2306.04938/) (2023)
- [Textureless Object Recognition: An Edge-based Approach](https://mlsanity.com/paper/2403.06107/) (2024)
- [Exploring Machine Learning Engineering for Object Detection and Tracking by Unmanned Aerial Vehicle (UAV)](https://mlsanity.com/paper/2412.15347/) (2024)
- [Comparative Analysis of Object Detection Algorithms for Surface Defect Detection](https://mlsanity.com/paper/2510.21811/) (2025)
- [Transformer-Driven Multimodal Fusion for Explainable Suspiciousness Estimation in Visual Surveillance](https://mlsanity.com/paper/2512.09311/) (2025)

## Abstract (arXiv)

Object detection has gained great progress driven by the development of deep learning. Compared with a widely studied task -- classification, generally speaking, object detection even need one or two orders of magnitude more FLOPs (floating point operations) in processing the inference task. To enable a practical application, it is essential to explore effective runtime and accuracy trade-off scheme. Recently, a growing number of studies are intended for object detection on resource constraint devices, such as YOLOv1, YOLOv2, SSD, MobileNetv2-SSDLite, whose accuracy on COCO test-dev detection results are yield to mAP around 22-25% (mAP-20-tier). On the contrary, very few studies discuss the computation and accuracy trade-off scheme for mAP-30-tier detection networks. In this paper, we illustrate the insights of why RetinaNet gives effective computation and accuracy trade-off for object detection and how to build a light-weight RetinaNet. We propose to only reduce FLOPs in computational intensive layers and keep other layer the same. Compared with most common way -- input image scaling for FLOPs-accuracy trade-off, the proposed solution shows a constantly better FLOPs-mAP trade-off line. Quantitatively, the proposed method result in 0.1% mAP improvement at 1.15x FLOPs reduction and 0.3% mAP improvement at 1.8x FLOPs reduction.