Yolo Object Detection in Google Colab [Full Tutorial]

not aspire is better so this is in this video I am talking about the Yolo object detection so we are using all Yolo v3 version so in this video I will show how you can implement Yolo in Google collab so normally you for we are trying to use the Yolo in our in our system in our local machine so it will require a higher GPU power so obviously we cannot afford a higher GPU power so we can do this thing on google call up also so let's see how we can do this thing in Google polyp so let's see before starting the video just I am going to tell you that the subscriber machine learning hub channel you will get a different types of machine learning computer vision OpenCV and different types of videos you can see already I operate flask with sleeping fashion development car detection MLM number plus recognization there are number of playlet you can see here so now let's focus on our videos just subscribe the channel so now we require storage for the storage we need to initialize Google Drive here so this is the code that you this notebook link in description you can copy the notebook from link so this URL will open in you where to paste this code here so it will connect to your Google Drive - with our Google column so our Google caller will get the storage so now we need to initialize the GPU so now you have to go to the runtime and change runtime type currently I am using the Google GPU so just you were to use the GPU for the you'll object detection so now run this part so it will show you information related to the GPU so we are using this GPU cuda version 10.1 so now let's go ahead now we need to clone this our darknet repository of Yolo Yolo v3 so import West via we just set the environment and there are some Linux and normally CMD commanded you need to know so we are get cloning our depository here so it will clone the repositories of this github so let's see so this repository will clone here so now let's go ahead now we need to install some GCC and most of the compiler based program which will allow us to do this yellow object detection on our Google column so these all are the commands so just run this command so guys the you will get all this command in this code also even this command is also available in this official repository so it will downloading some needed file from the Ubuntu servers so it will take some time to download these files so guys these this all the procedure will take around almost a one-in-four also because it will take a time so let us see now if we need to initialize the darknet folder and compile the GPU so we are initializing our GPU here so let first we need to change the word a working directory into dark red folder or darknet is our clone the repository from the g-tube so it will take around 3 to 4 minutes depends on the your internet speed and colab actually in Kolhapur internet speed no issue because they provide we actually connect to the Google server the speed would be awesome for all the all the users the most time-consuming process is this we need to download the you load III wait wait file so this actually this sidecar uber is slow I don't know why my internet speed is it to Google but actually this is taking it too much time to download this wait file so it will take it too much time after this getting the wait you don't need to do anything all the things done just we are to upload our video means testing video it will process the video then after that we can download the play process video so it will take a time but most of the time consuming process is the downloading wait from the official site because site is a very slow okay so this procedure is completed now it's time to get the weight so after getting the weight we are changing the permissions of the darknet folder so let's download the weight and it will take too much time around half train for maximum happened over more than even one or also possible you can see estimated time waits 45 minute so currently I have a 50 Mbps plan but I don't know what happening here actually the service is low so I am pushing with you here so wait file is downloaded iturra it around takes a 25 minutes to download so now we change need to change it our working directory so we are using PWD command so present working directory so now our Yolo v3 properly initialized now just we have to sorry my mic is not working I think so just now we have to upload our file so before uploading file we need to install some of the this kind this library so here command is defined so you can see almost all of the required libraries defined here and so with this command this all library will be installed so after library installation we will upload the video here so before this procedure just already already one videos which is shorter than around one minute or two minute so it takes this libraries to written takes around two to three minutes okay so now relief installation is totally complex so now it's time to the upload of the video so we're this is the Google column inbuilt every from google.com up import files so it will provide a file uploader button here so choose file you just throw download I have downloaded this one file paste file so I will upload this file here so it will take some time to approve so guys my video file is now uploaded so in this command we have to define our video file so already all the required file is already defined here so just we need to define over a file here here we defined don't show because if we depend don't show then it will not show the processing video it will do background processing so here we need to define our video so my video's name is your taste know 10 people then we need to define a taste note mp4 now we need to define output file name is equal to any file name but this file this polish extension should be avi format so my I define output 1 dot avi so now it's time to run this process I run this cell so it will use the GPU power most so you can see convolution neural network is now just started fitting and you can see car person busker so it will pre process the video after calm process you're completing we can download the video so you can see this is this doing procedure on the one by one frames so after this procedure completion we will get the video so you can see power of the GPU this video is a full in 1 1080p format but it takes around or less than a 30 second so now I can download the video here so this is the command from google koala import file file store download output dot avi because our working directory is this so we can download this output dot avi file from here so it will take much around 2-3 minute to download the file so let's see how output will look so my video is now downloaded so just let see how it looking so this is a actually video you can see this is the security footage of the dildo goes down highway so you can see every let's pause the video so you can see every percentage of detector every everything this detector you can see this person this car even this person is also detected these all are the road workers here detected you can see this car you want this tourist bus is also detected you can see this biker is also detector I am playing once again so guys they're all for today video you can see how you'll object detection is working so I will make one other video on the custom object detection means I will detect the custom object with the Yolo object detection code so liked this video share with your friend and don't forget to subscribe the machine learning of Channel so see you in the next one

Channel: Machine Learning Hub

Views: 48,102

Keywords: yolo object detection, yolo object detection tutorial, yolo object detection python, google colab, train yolo on google colab, yolo on google, yolo on google colab, yolo on google cloud, yolo, yolov3 tutorial, yolov4, yolov3, object detection yolo, object detection yolov3, machine learning hub, object detection

Id: bTIKY2NYBUg

Length: 10min 54sec (654 seconds)

Published: Wed Jun 24 2020