Code Monkey home page Code Monkey logo

faster-rcnn-pedestrian-detection's Introduction

Pedestrian-RCNN

This code extends py-faster-rcnn by adding:

  • ResNet implementation.
  • Online Hard Example Mining.
  • Caltech Dataset train and test interface.
  • Add RPN tools.

Base

Installation

The installation and useage are same as Faster R-CNN

  1. clone the repository and caffe submodule
  2. build lib and caffe
  3. train your model command like: ./experiments/scripts/faster_rcnn_end2end.sh 0 ResNet50 caltech_reasonable

Notices

For different datasets, the interval frame rate is different, so it need to be changed.

In ./lib/datasets/caltech.py, change the start_frame and frame_rate of insert_frame function.

faster-rcnn-pedestrian-detection's People

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar

faster-rcnn-pedestrian-detection's Issues

关于merged版本

@chaopei HI

在您的models下,您提供了Merged版本和非Merged版本.根据对应的prototxt文件,感觉Merged版本就是用一个Scale层代替了BatchNorm层+Scale层,非Merged版本就是传统的BatchNorm层+Scale层

OHEM on VGG

hello, thank you for sharing the repo. Here I have two questions:

  1. did you test the ohem on vgg, and what about the result?
  2. what does VGG_CONCAT mean?

Where do you get trainval.txt?

Hello,

When I run
bash ./experiments/scripts/caltech_end2end.sh 0 ResNet50 caltech_reasonable,
I get the following output
Path does not exist: data/caltech-pedestrian-dataset-converter/data/trainval.txt

caltech-pedestrian-dataset-converter only generate image and annotations. Where do you get trainval.txt?

caltech detection result on reasonable

Hi thank you for you sharing~

Could you please show your miss rate of test set in reasonable condition?

l used py-faster-rcnn VGG16 on this test set, but the detection result is not good , the mAP=0.35
l don't know the reason?

could you give me some suggestions for solving the problem

about BN_Scale_Merged

@chaopei HI

在models下,您提供了Merged版本和非Merged版本.根据对应的prototxt文件,感觉

  1. Merged版本就是用一个Scale层代替了BatchNorm层+Scale层
  2. 非Merged版本就是传统的BatchNorm层+Scale层

想咨询下,

  1. 上述自己的理解对吗?
  2. 目前哪个版本的caffe支持融合BatchNorm层+Scale层的操作呢??即Merged版本
  3. 为什么基于同一个caffe版本,既支持传统的BatchNorm层+Scale层操作,又支持用一个Scale层代替了BatchNorm层+Scale层的操作呢?

about BN_Scale_Merged

@chaopei HI

在models下,您提供了Merged版本和非Merged版本.根据对应的prototxt文件,感觉
1. Merged版本就是用一个Scale层代替了BatchNorm层+Scale层
2. 非Merged版本就是传统的BatchNorm层+Scale层

想咨询下,
1. 上述自己的理解对吗?
2. 目前哪个版本的caffe支持融合BatchNorm层+Scale层的操作呢??即Merged版本
3. 为什么基于同一个caffe版本,既支持传统的BatchNorm层+Scale层操作,又支持用一个Scale层代替了BatchNorm层+Scale层的操作呢?

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.