tf-faster-rcnn | Tensorflow Faster RCNN for Object Detection | Machine Learning library

by endernewton Python Version: v0.12 License: MIT

X-Ray Key Features Code Snippets(3)Community Discussions(2)Vulnerabilities Install Support

kandi X-RAY | tf-faster-rcnn Summary

tf-faster-rcnn is a Python library typically used in Artificial Intelligence, Machine Learning, Tensorflow applications. tf-faster-rcnn has no bugs, it has no vulnerabilities, it has a Permissive License and it has medium support. However tf-faster-rcnn build file is not available. You can download it from GitHub.

Tensorflow Faster RCNN for Object Detection

Support

Quality

Security

License

Reuse

Support

tf-faster-rcnn has a medium active ecosystem.

It has 3635 star(s) with 1612 fork(s). There are 119 watchers for this library.

It had no major release in the last 12 months.

There are 211 open issues and 270 have been closed. On average issues are closed in 129 days. There are 5 open pull requests and 0 closed requests.

It has a neutral sentiment in the developer community.

The latest version of tf-faster-rcnn is v0.12

Quality

tf-faster-rcnn has 0 bugs and 0 code smells.

Security

tf-faster-rcnn has no vulnerabilities reported, and its dependent libraries have no vulnerabilities reported.

tf-faster-rcnn code analysis shows 0 unresolved vulnerabilities.

There are 0 security hotspots that need review.

License

tf-faster-rcnn is licensed under the MIT License. This license is Permissive.

Permissive licenses have the least restrictions, and you can use them in most projects.

Reuse

tf-faster-rcnn releases are available to install and integrate.

tf-faster-rcnn has no build file. You will be need to create the build yourself to build the component from source.

Installation instructions, examples and code snippets are available.

Top functions reviewed by kandi - BETA

kandi has reviewed tf-faster-rcnn and discovered the below as its top functions. This is intended to give you an instant insight into tf-faster-rcnn implemented functionality, and help decide if they suit your requirements.

Find the target anchor layer
Transform a bounding box
Unmap data
Compute the targets according to the input data
Create the architecture
Add the activity summary
Add the image to the image
Adds a summary for the background image
Convert from recommre to TRA
Creates the architecture
Draw bounding boxes
Return a list of gt roidb
Generate pre - generated anchors
Train a network
Create detections from a given IMDB file
Evaluate the detection results
Fix vGG16 variables
Create a combined ROI model
Append flipped images
Visualize object classes
Create a config dictionary from a list
Generate proposal target layer
Parse command line arguments
Locate CUDA
Convert resnet to head
Generate proposal top layer
Convert image to head

Get all kandi verified functions for this library.

tf-faster-rcnn Key Features

No Key Features are available at this moment for tf-faster-rcnn.

tf-faster-rcnn Examples and Code Snippets

4 Train your own model on other dataset,4.2 Modify the program

Jupyter Notebook

Lines of Code : 32

License : No License

Copy

def _get_widths(self):
    return [PIL.Image.open(self.image_path_at(i)).size[0]
            for i in range(self.num_images)]

def _get_heights(self):
    return [PIL.Image.open(self.image_path_at(i)).size[1]
            for i in range(self.num_image

pytorch-faster-rcnn,特别说明

Python

Lines of Code : 25

License : Permissive (MIT)

Copy

# 为了融合全局特征，在roi pooling前加了类似U-Net的东西
ZDF_GAUSSIAN: False
ZDF: True
# 在原有分类基础上加了细分类，目的是通过multi-task提升原有的分类、检测和mask
SUB_CATEGORY: False
LOSS_SUB_CATEGORY_W: 0.5
# 这两个参数应对不同的POOLING_MODE
# pyramid_crop_sum金字塔roi(1,1.5,2)
# pyramid_crop金字塔roi cat后降维
# 其他

3 Train your own model on the VOC dataset,3.2 Modify shell file

Jupyter Notebook

Lines of Code : 15

License : No License

Copy

pascal_voc)
    TRAIN_IMDB="voc_2007_trainval"
    TEST_IMDB="voc_2007_test"
    STEPSIZE="[50000]"
    ITERS=70000 ---> 300 
    ANCHORS="[8,16,32]"
    RATIOS="[0.5,1,2]"
    ;;

pascal_voc)
    TRAIN_IMDB="voc_2007_trainval"
    TEST_IMDB="voc_

Community Discussions

Trending Discussions on tf-faster-rcnn

In FasterRCNN, why is the short edge size 600 and the max size 1000?

How to reduce size of bottleneck features of intermediate layer of VGG16?

QUESTION

In FasterRCNN, why is the short edge size 600 and the max size 1000?

Asked 2019-Feb-01 at 07:59

I read the following code:

https://github.com/endernewton/tf-faster-rcnn/blob/a3279943cbe6b880be34b53329a4fe3f971c2c37/lib/model/config.py#L63

600 is the pixel size of an image's shortest side, and 1000 is the max pixel size of the longest side of a scaled input image. Could anybody explain this? and how to determine these sizes? Shall we change these sizes?

...

ANSWER

Answered 2019-Feb-01 at 07:59

These are used in prep_im_for_blob function in here. Where target_size is __C.TRAIN.SCALES = (600,), and max_size is __C.TRAIN.MAX_SIZE = 1000. What it does is scales the image so that the minimum size of the resized image is equal to __C.TRAIN.SCALES. However if the resulting image becomes bigger than __C.TRAIN.MAX_SIZE it scales so that maximum size of resized image is equal to __C.TRAIN.MAX_SIZE. If your input image typically falls within 600~1000 pixels in range, you don't need to change these values.

Source https://stackoverflow.com/questions/54458086

QUESTION

How to reduce size of bottleneck features of intermediate layer of VGG16?

Asked 2017-Sep-06 at 13:14

I am trying to connect the layer c0nv4_3 of vgg16 network instead of conv5_3 to the RPN network of Faster R-CNN. Here is the python code of vgg16 network. I have changed these lines:

...

ANSWER

Answered 2017-Sep-05 at 10:02

There are methods too, for reducing the length of bottleneck features.

Why not to add deconv:

You will initialize deconv layer with random values
You are not finetuning the network, you are just making forward pass through the network.
So the output of deconv will randomize your features of conv4

Pooling Layers:

Average pooling(based on the window size, it will return average of that window). So if lets say window(2,2) with values[3,2,4,3] will result into only one value: 6
MaxPool(based on the window size, it will result maximum value of that window). So if lets say window(2,2) with values[3,2,4,3] will result into only one value: 3

Check out pooling layers here

Source https://stackoverflow.com/questions/45944032

Community Discussions, Code Snippets contain sources that include Stack Exchange Network

Vulnerabilities

No vulnerabilities reported

Install tf-faster-rcnn

Grid K520 (AWS g2.2xlarge). Tesla K80 (AWS p2.xlarge). Note: You are welcome to contribute the settings on your end if you have made the code work properly on other GPUs. Also even if you are only using CPU tensorflow, GPU based code (for NMS) will be used by default, so please set USE_GPU_NMS False to get the correct output.
Clone the repository
Update your -arch in setup script to match your GPU
Build the Cython modules
Install the Python COCO API. The code requires the API to access COCO dataset.
Please follow the instructions of py-faster-rcnn here to setup VOC and COCO datasets (Part of COCO is done). The steps involve downloading data and optionally creating soft links in the data folder. Since faster RCNN does not rely on pre-computed proposals, it is safe to ignore the steps that setup proposals.

Support

For any new features, suggestions and bugs create an issue on GitHub. If you have any questions check and ask questions on community page Stack Overflow .

Find more information at: