论文阅读-《BlitzNet: A Real-Time Deep Network for Scene Understanding》
2017-08-15 15:52
746 查看
ICCV 2017
采用的是Resnet50+SSD, ssd这种one-stage的检测器天生适合和分割一块做。上采样过程用到的block如下图所示，除了正常的skip connection之外，还用上了residual connection
这里就放一张在VOC2007上的测试结果。因为加上了分割的标签，所以检测的效果肯定会更好。
上图表示在VOC2012验证集上的结果，可以看到分割和检测这两个任务可以相互促进
1.Motivation:
为了做到实时的目标检测和语义分割2.Framework
采用的是Resnet50+SSD, ssd这种one-stage的检测器天生适合和分割一块做。上采样过程用到的block如下图所示，除了正常的skip connection之外，还用上了residual connection
3.Experiments
作者在VOC2007/2012以及COCO上做的实验，因为VOC上不是每一张有boungbing box标注的图像都有segmentation的标注，因此作者还用到了别人另外标注的分割标签：《Semantic contours from inverse detectors》这里就放一张在VOC2007上的测试结果。因为加上了分割的标签，所以检测的效果肯定会更好。
上图表示在VOC2012验证集上的结果，可以看到分割和检测这两个任务可以相互促进
内容来自用户分享和网络整理，不保证内容的准确性，如有侵权内容，可联系管理员处理 
相关文章推荐
- 目标检测分割--BlitzNet: A Real-Time Deep Network for Scene Understanding
- 【那些年我们一起看过的论文】之《ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation》
- 论文阅读之：Deep Meta Learning for Real-Time Visual Tracking based on Target-Specific Feature Space
- 论文阅读--PVANET: Deep but Lightweight Neural Networks for Real-time Object Detection
- Multi-style Generative Network for Real-time Transfer论文理解
- 论文阅读（Weilin Huang——【TIP2016】Text-Attentional Convolutional Neural Network for Scene Text Detection）
- 论文阅读：Deep MANTA: A Coarse-to-fine Many-Task Network for joint 2D and 3D vehicle analysis
- Semantic Segmentation--ENet:A Deep Neural Network Architecture for Real-Time Semantic..论文解读
- 论文阅读-Re 3 : Real-Time Recurrent Regression Networks for Object Tracking
- A Twofold Siamese Network for Real-Time Object Tracking 阅读笔记
- READING NOTE: ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation
- 多尺度R-CNN论文笔记(4): PVANET: Deep but Lightweight Neural Networks for Real-time Object Detection
- SCNN车道线检测--(SCNN)Spatial As Deep: Spatial CNN for Traffic Scene Understanding（论文解读）
- 【论文阅读】Fused Text Segmentation Network for Multi-oriented Scene Text Detection
- 论文阅读（Xiang Bai——【PAMI2017】An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition）
- 论文笔记：PVANET: Deep but Lightweight Neural Networks for Real-time Object Detection
- PVANET----Deep but Lightweight Neural Networks for Real-time Object Detection论文记录
- 13.5 论文笔记：目标追踪-CVPR2014-Adaptive Color Attributes for Real-time Visual Tracking
- 论文阅读理解 - Deep Learning of Binary Hash Codes for Fast Image Retrieval
- 论文阅读笔记 Picking Deep Filter Responses for Fine-grained Image Recognition