This extension provides tools to easily collect depth and color data from a top view of cattle
Picturecapture, Videocapture, and ConvertToPNG can only be run on a Windows platform. ThresholdExtract can be run on Windows or Mac.
- python (any version 3.6 to 3.9)
- code editor (Visual Studio Code is what I recommend)
- Realsense SDK https://www.intelrealsense.com/sdk-2/
- Realsense SDK install instructions for Windows https://www.youtube.com/watch?v=EOJHsNIayio&ab_channel=IntelRealSense
- python (any version since 3.6 to 3.9)
- code editor (Visual Studio Code is what I reccomend)
- opencv-python
- time
- datetime
- keyboard
- numpy
- tkinter
- opencv-python
- keyboard
- numpy
- imutils
pip install CattleRealsense==0.1.7
from CattleRealsense.PictureCapture import PictureCapture as ps
#Instance of a PictureCapture
pictureTake = ps()
#location to save pictures
#make sure to convert all \ to \\
#example, if you would like to save the picutres to a folder in your desktop, saveLocaitonPath will look something like
saveLocationPath = "C:\\Users\\johndoe\\Desktop\\BAG\\" #files will be saved in a folder named BAG
#Lets user set the point of interest, depth, number of pictures, and time in between pictures
#picture collection will start after setup is complete
pictureTake.setupManual(saveLocationPath)
When this is running you will be shown a depth and color picture from your camera. Click a location on either picture to set a point of interest. After clicking type r to refresh the image. When satisfied with point of interest location hit enter on your keyboard.
to use the same setting as used by Virginia Tech (x = 500, y = 360, distance_to_ground = 3 meters, num_of_picutres = 10, time_between_pics = 0.15s save_location = foulder where code is)
from CattleRealsense.PictureCapture import PictureCapture as ps
pictureTake = ps()
pictureTake.setupAuto(saveLocationPath)
to run custom collection without going through the setup process
from CattleRealsense.PictureCapture import PictureCapture as ps
#Instance of a PictureCapture
pictureTake = ps()
#fill in parameters to match desired
pictureTake.runCamera(dist_to_ground, xCord, yCord, number_of_pics, sec_between_pics, save_location)
All recording processes will stop when user ends program
this is the same process as Picture Collection but without number_of_pics, sec_between_pics
from CattleRealsense.VideoCapture import VideoCapture as vc
#location to save videos
#make sure to convert all \ to \\
#example, if you would like to save the picutres to a folder in your desktop, saveLocaitonPath will look something like
saveLocationPath = "C:\\Users\\johndoe\\Desktop\\BAG\\" #files will be saved in a folder named BAG
videoTake = vc()
#to setup manually
videoTake.setupManual(saveLocationPath)
to use the same setting as used by Virginia Tech
from CattleRealsense.PictureCapture import PictureCapture as ps
#Instance of a PictureCapture
videoTake = vc()
videoTake.setupAuto(saveLocationPath)
to run collection without giong through the setup process
from CattleRealsense import PictureCapture as ps
#Instance of a PictureCapture
videoTake = vc()
#fill in parameters to match desired
videoTake.runCamera(self, dist_to_ground, xCord, yCord, save_loc)
During data collection all files will be saved as .bag. This makes it hard to use. To convert the .bag files to .png follow the following code
from CattleRealsense.ConvertToPNG import ConvertToPNG as convert
con = convert()
#inputFolder = Path to .bag folder
#outputFolder = path to save .png created
#toolsPath = path where \\Intel RealSense SDK 2.0\\tools\ is saved
#toolspath will likely be "C:\\Program Files (x86)\\Intel RealSense SDK 2.0\\tools\\"
#run the conversion program
con.convert(inputFolder, outputFolder, toolsPath)
To get an accurate threshold it is important to crop the image horizontally to the smallest area needed. This prevents the surrounding objects from interfering. When running this program the first picture in from inputFolder will be shown.
- You should click the upper bound, lower bound, left bound, right bound in that order.
- After clicking the mouse 4 times hit enter on your keyboard
- The cropped photo will appear. If you are happy with the result hit enter again to run thresholding.
- If you would like to select new crop values type r
from CattleRealsense.ThresholdExtract import ThresholdExtract as te
extract = te()
inputFolder = path to .png files
outputFolder = path to save threshold results
extract.setCrop(inputFolder, outputFolder)
To prevent having to setup the crop every time you may input the parameters directly.
from CattleRealsense.ThresholdExtract import ThresholdExtract as te
inputFolder = path to .png files
outputFolder = path to save threshold results
extract = te()
extract.extract(y1Set, y2Set, x1set , x2set, inputFolder, outputFolder)
#y1Set = Upper bound y
#y2Set = Lower bound y
#x1Set = Left bound
#x2Set = Right bound
#inputFolder = folder to read from
#outputFolder = folder to output to
Follow these instructions if you would like to use Visual Studio Code (VSC)
- download VSC
- open VSC click file --> new file
- click 'select a language' and chose python
- type command s and enter desired name followed by .py
- type
print('Hello World')into the file. save the file - on the top toolbar select Run --> Run Without Debugging
- if you are asked to install an extension do so. The top option from the search bar should work fine. run the program again
- if 'Hello World' is printed in the terminal you are good to continue.
In the terminal type python3 -m pip install CattleRealsense==0.1.7 or sudo python3 -m pip install CattleRealsense==0.1.7.
Make sure to pip install all the required packages listed at the top of this page.
To get an accurate threshold it is important to crop the image horizontally to the smallest area needed. This prevents the surrounding objects from interfering. When running this program the first picture in from inputFolder will be shown.
- You should click the upper bound, lower bound, left bound, right bound in that order.
- After clicking the mouse 4 times hit enter on your keyboard
- The cropped photo will appear. If you are happy with the result hit enter again to run thresholding.
- If you would like to select new crop values type r
from CattleRealsense.ThresholdExtract import ThresholdExtract as te
extract = te()
#inputFolder = saveLocationPathMac = "X: /Users/username/Desktop" #if files are on desktop
#outputFolder = path to save threshold results
extract.setCrop(inputFolder, outputFolder)
You must run the script as an admin
enter this into the terminal to run the script
sudo python3 -m name.py Where name is the name of the file you have your work in.
Make sure your terminal is in the directory where you saved your file.
To prevent having to setup the crop every time you may input the parameters directly.
from CattleRealsense.ThresholdExtract import ThresholdExtract as te
inputFolder = path to .png files
outputFolder = path to save threshold results
extract = te()
extract.extract(y1Set, y2Set, x1set , x2set, inputFolder, outputFolder)
#y1Set = Upper bound y
#y2Set = Lower bound y
#x1Set = Left bound
#x2Set = Right bound
#inputFolder = folder to read from
#outputFolder = folder to output to
You must run the script as an admin
enter this into the terminal to run the script
sudo python3 -m name.py Where name is the name of the file you have your work in.
Make sure your terminal is in the directory where you saved your file.