视觉传感器:2D感知算法
来源 | 巫婆塔里的工程师@知乎
1 前言
2 物体检测
2.1 两阶段检测
2.2 单阶段检测
2.3 无Anchor检测
2.4 性能对比
3 物体跟踪
-
由物体检测器在单帧图像上得到物体框输出。 -
提取每个检测物体的特征,通常包括视觉特征和运动特征。 -
根据特征计算来自相邻帧的物体检测之间的相似度,以判断其来自同一个目标的概率。 -
将相邻帧的物体检测进行匹配,给来自同一个目标的物体分配相同的ID。
4 语义分割
参考文献:
[1] Girshick et al., Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation, 2014.
[2] Girshick, Fast R-CNN, 2015.
[3] Ren et al., Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, 2016.
[4] Lin et al., Feature Pyramid Networks for Object Detection, 2017.
[5] Liu et al., SSD: Single Shot MultiBox Detector, 2015.
[6] Lin et al., Focal Loss for Dense Object Detection, 2017.
[7] Redmon et al., You Only Look Once: Unified, Real-Time Object Detection, 2015.
[8] Zhou et al., Objects as Points, 2019.
[9] Law and Deng, CornerNet: Detecting Objects as Paired Keypoints, 2019.
[10] Yang et al., RepPoints: Point Set Representation for Object Detection, 2019.
[11] Carion et al., End-to-End Object Detection with Transformers, 2020.
[12] Ciaparrone et al., Deep Learning in Video Multi-Object Tracking: A Survey, 2019.
[13] Bewley et al., Simple Online and Realtime Tracking, 2016.
[14] Wojke et al., Simple Online and Realtime Tracking with A Deep Association Metric, 2017.
[15] Milan et al., Online Multi-Target Tracking using Recurrent Neural Networks, 2017.
[16] Zhou et al., Tracking Objects as Points, 2020.
[17] Long et al., Fully Convolutional Networks for Semantic Segmentation, 2014.
[18] Chen et al., DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs, 2017.
[19] Peng et al., Large Kernel Matters —— Improve Semantic Segmentation by Global Convolutional Network, 2017.