YOLO-V3 for Windows

What is YOLO-V3?

YOLO ,or You Only Look Once is a object detion algoritham. iT's the most accurate real time object detection algorithm. If want to more about YOLO-V3 then you check here.

So recently i was playing with yolo-v3 (i was also do little bit yolo-v2) and since now a days PUBG moblie is very famous to i tried this yolo on pubg.Though i don't get high accuracy but it's ok because i was trained this model just around 300 hundred of images(i was too much excited about that) but now i know little about the procedure so again i will do with more images.

This madole is trained on 4 classes : car,Medicine,Weapon,Enwmies.

This is my model demo(Video demo)

If you want to use my model then follow the steps ,i make it very simple because i was (still) very complicated for me so i try to break is down. Here Extra you have nothing to do extra.So jump right into it...

How You Can Use My model?

Step-1:

Download:

case-1:

Download the zip of this repo and extract it. And notice that after dowload file format should be this

In this case after extracting you will see a bin folder you have to go their D:\YOLO-V3\bin this bin folder and extract the zip file(to get the full weights).

(like shown in image just extract that)

Note

After that go to: data folder and extract lebels.zip .

Case-2:

Or you can download all the file from my google drive here

Now if think that so why i upload this file file on github ? simple answer is Just for looks good😂.

Ok so next move on to step-2.

step-2:

Download OpenCV from here

automatically downlaod will start .

now guys here little thing to do

1.

Go to C:\ if you see any version of openCV then remane it to some thing else or delete it (easy one😁).

2.

then extract the downloaded OpenCV in C:/ like this below image

3.

After extracting if openCV has name some thing else rather than this imagen one ,then just rename and make it this below image one and the path of open should be this : C:\opencv_3.0\opencv .

Step-3:

After step-2 go to : C:\opencv_3.0\opencv\build\x64\vc14\bin this path ,copy it and add this path to your Environment Variables. If you don't do step-4 you will get an error opencv_world330d.dll was not found something like that so do it.

Step-4:

Open cmd under YOLO-V3 folder and run this command :darknet_no_gpu.exe detect cfg/yolov3_tiny_pubg.cfg bin/yolov3_tiny_pubg_2000.weights data/pubgcar.jpg if you follow the above steps then it should work and it will pop up a predicted image. Congrats you successfully run YOLO-V3.

Explaination of this code:

darknet_no_gpu.exe: Running darknet no GPU version. detect: This command is for process images for video code is demo we will do it later. cfg/yolov3_tiny_pubg.cfg : Path of cfg file. bin/yolov3_tiny_pubg_2000.weights: Path of weights. data/pubgcar.jpg: path of the image.

YOLO on video file

just run this command: darknet_no_gpu.exe detector demo cfg/coco.data cfg/yolov3_tiny_pubg.cfg bin/yolov3_tiny_pubg_2000.weights -dont_show videofile_name.mp4 -i 0 -out_filename result.avi (don't use more than 1 or 2 min video because if you no gpu then its gonna take to much time to process it)

your video file will be saved in result.avi

enjoy feel free to try your own images.