Download presentation
Presentation is loading. Please wait.
Published byNicolas de Santarém Modified 5年之前
1
Research 裴澍炜 Shuwei Pei Tel:
2
Two projects 1:基于视频的多目标检测 detection based on video
2:基于视频的多目标跟踪 track based on video Target: bike,pedestrian and so on. (which related with autonomous vehicles).
3
视频的多目标检测 ( detection ) 输入图像 Input Algorithm 输出标注 Output 评估算法 Evaluate
研究基于视频(图像)的目标检测算法,编写相应的代码应用。 Using deep learning algorithm. Based on caffe (a open source project)frame. . 以行人检测为例(方法必须为深度学习的方法): 基于caffe深度学习架构 输入图像 Input Algorithm 输出标注 Output 评估算法 Evaluate
4
基于视频的行人检测 (detection track)
第一个项目检测:What need to do 基于caffe深度学习架构(RCNN) 输入图像 输出标注 评估算法 1、批量图像的输入,各个模块间的接口设计,可选语言:matlab/python Interface design 2、模型的训练、测试过程 The key point is to design the model or optimize the model. 3、采用新的模型进行训练、测试。Train and test.
5
基于视频的多目标跟踪 ( track ) 与检测类似,实现多目标跟踪算法。
Like detection, implements of the algorithm. 输入为相关标准数据集,评估算法性能。 You should evaluate your resault. I am now studying the first project(detection). So I may not give too much advices about the second project(track) .
Similar presentations