Venue
European Conference on Computer Vision
We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For 300 × 300 input, SSD achieves 74.3% mAP 1 on VOC2007 test at 59 FPS on a Nvidia Titan X and for 512 × 512 input, SSD achieves 76.9% mAP, outperforming a comparable state-of-the-art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at: https://github.com/weiliu89/caffe/tree/ssd .
The paper introduces SSD (Single Shot MultiBox Detector), a method for real-time object detection using a single convolutional neural network. SSD simplifies object detection by eliminating the proposal generation step used in previous methods, instead predicting class scores and bounding box adjustments for a fixed set of default bounding boxes across multiple feature maps. Evaluated on datasets such as PASCAL VOC, COCO, and ILSVRC, SSD demonstrates competitive accuracy and improved speed compared to conventional methods like Faster R-CNN and YOLO. Key contributions include efficient handling of various object sizes through multi-scale feature maps and an innovative training approach that includes hard negative mining and extensive data augmentation, achieving high accuracy even with low-resolution inputs.
This paper employs the following methods:
- Convolutional Neural Network
The following datasets were used in this research:
- SSD achieves 74.3% mAP on VOC2007 test at 59 FPS
- SSD512 outperforms Faster R-CNN by 1.7% mAP
The authors identified the following limitations:
- Sensitivity to bounding box sizes, especially smaller objects
- Potential for further improvements in default box tiling for better detection performance on small objects
- Number of GPUs: 1
- GPU Type: Nvidia Titan X
Single Shot MultiBox Detector
SSD
real-time detection
convolutional neural network
multi-scale detection