YOLOv3: An Incremental Improvement
By Joseph Redmon, Ali Farhadi. 2018
25,668 citations recorded by the source corpus.
We present some updates to YOLO! We made a bunch of little design changes to make it better. We also trained this new network that's pretty swell. It's a little bigger than last time but more accurate. It's still fast though, don't worry. At 320x320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3 is quite good. It achieves 57.9 mAP@50 in 51 ms on a Titan X, compared to 57.5 mAP@50 in 198 ms by RetinaNet, similar performance but 3.8x faster. As always, all the code is online at this https URL
Subfields: Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Xiaogang Wang
- Trevor Darrell
- Yi Yang
- Jitendra Malik
- Chunhua Shen
In Radiology, Nuclear Medicine and Imaging
- Hongpeng Liao
- Chien-Yao Wang
- Diana-Margarita Córdova-Esparza
- Juan R. Terven
- Alexey Bochkovskiy
- Diganta Misra
- Quoc V. Le
- I-Hau Yeh
Citations in the corpus
- Strike (With) a Pose: Neural Networks Are Easily Fooled by Strange Poses of Familiar Objects (2018, methodology): “Here, we investigate how well our AXs transfer to a state-of-the-art object detector— YOLOv3.”
- Objects as Points (2019, methodology): “With flip testing, our model is still faster than YOLOv3 [45] and achieves accuracy levels of Faster-RCNN-FPN [46] (CenterNet 39 .”
- FCOS: Fully Convolutional One-Stage Object Detection (2019, background): “All current mainstream detectors such as Faster R-CNN [24], SSD [18] and YOLOv2, v3 [23] rely on a set of pre-defined anchor boxes and it has long been believed that the use of anchor boxes is the key to detectors’ success.”
- Generalized Intersection Over Union: A Metric and a Loss for Bounding Box Regression (2019, methodology): “Most popular object detectors [20, 21, 3, 12, 13, 16] utilize some combination of the bounding box representations and losses mentioned above.”
- Instance Segmentation for Autonomous Log Grasping in Forestry Operations (2022, background): “[25] published a dataset of standing trees for bounding boxes detection and trained a YOLOv3 [26] network on it.”
- Towards Real-World Prohibited Item Detection: A Large-Scale X-ray Benchmark (2021, methodology): “After that, different one-stage object detectors are proposed, including SSD [24], DSSD [8], and YOLO series [28, 29, 30].”
- Domain Adaptation for Traffic Density Estimation (2021, background): “With the advent of Convolutional Neural Networks (CNNs) (Lecun et al., 1998), supervised learning has reached excellent results across many Computer Vision application areas, such as object detection (Redmon and Farhadi, 2018) and instance segmentation (He et al., 2017).”
- Real-World Image Datasets for Federated Learning (2019, methodology): “With the rapid development of deep learning, object detection algorithms have been greatly improved in the past few decades [ Ren et al. , 2017; Redmon and Farhadi, 2018a; Zhao et al. , 2018 ] .”
- EfficientDet: Scalable and Efficient Object Detection (2019, methodology): “32.4 2.5B YOLOv3[26] 33.0 71B(28x) EfficientDet-D1 38.3 6.0B RetinaNet[17] 37.0 97B(16x) MaskRCNN[8] 37.9 149B(25x) EfficientDet-D5 ‡ 49.8 136B AmoebaNet+NAS-FPN[37] ‡ 48.6 1317B(9.7x) EfficientDet-D7 † 51.0 326B AmoebaNet+NAS-FPN[37] †‡ 50.7 3045B(9.3x) † Notplotted, ‡ trainedwithauto-augmentation[37]”
- Multi-Object Representation Learning with Iterative Variational Inference (2019, background): “…results on real-world images, however, many approaches (such as “semantic seg-mentation” or object detection) rely on supervised signals (Girshick, 2015; He et al., 2017; Redmon & Farhadi, 2018), while others require hand-engineered features (Shi & Malik, 2000; Felzenszwalb & Huttenlocher, 2004).”
- EPicker is an exemplar-based continual learning approach for knowledge accumulation in cryoEM particle picking (2022, methodology): “Some previous works dealing with sparse annotation in natural images solved this problem by reweighting the importance of region proposals generated by the two-stage object detector Faster RCNN 38,39 or recalibrating the loss of the anchors used by the one-stage object detector YOLO 40 .”
- Incremental Few-Shot Object Detection (2020, neutral): “Despite the success of deep convolutional neural networks (CNNs) [21, 24, 45, 49] in object detection [43, 42, 30, 28], most existing models can be only trained offline via a lengthy process of many iterations in a batch setting.”
- A Survey of Autonomous Driving: Common Practices and Emerging Technologies (2019, methodology): “More recent versions of this method, YOLOv2, YOLO9000 [137] and YOLOv3 [123] briefly took over the PASCAL VOC and MS COCO benchmarks while maintaining low computation and memory cost.”
- IQA: Visual Question Answering in Interactive Environments (2017, methodology): “ planning [73] and task oriented language grounding [8,19]. We extend this for the purpose of question answering. Since HIMN has access to object detections provided by either the environment or YOLO [56], we also provide detections to the baseline. For the baseline model, at each time-step, the raw”
- Few-Shot Object Detection via Feature Reweighting (2018, background): “The recent success of deep convolutional neural networks (CNNs) in object detection [32, 15, 30, 31] relies heavily on a huge amount of training data with accurate bounding box annotations.”
- YOLACT: Real-Time Instance Segmentation (2019, methodology): “MS that only has a marginal performance penalty. 1. Introduction “Boxes are stupid anyway though, I’m probably a true believer in masks except I can’t get YOLO to learn them.” – Joseph Redmon, YOLOv3 [36] What would it take to create a real-time instance segmentation algorithm? Over the past few yea”