Display label correctly

[hyunkyung12]

I want to display the label in my box.
However, it is displayed in black.
How can I display it?
What files or folders should I change?

[hyunkyung12]

I did it ! thank you!

[deepkshikha]

#hyunkyung12 what have you done to display the label what file you changed??

[hyunkyung12]

https://github.com/pjreddie/darknet/tree/master/data/labels
Copy the png files here and paste it into the path where the error occurred.
For example, if there is an error that there is no png file in the data folder, paste it into the data folder.

[deepkshikha]

You are running easy-yolo or other one. I am not understanding why should paste .png in data folder?

[hyunkyung12]

easy yolo was created to reduce train time in yolo.
I think the creator of easy yolo forgot to copy the data / labels folder while creating a new repository based on yolo's file.
Also, if you look at the source code, you have to specify the location of the label data. Actually, there is no such directory in the easy-yolo repository.
So if you create a folder called data / labels directly, the label will be displayed normally.

[deepkshikha]

Thank you for response. I have seen that in result it displayed couldn't upload images data/label/36_1.png something. So If I make a folder data/label and paste all .png file then Will testing occur over whole folder. And in which file of source code you have specify location of label data ??? In yolo.c ??

…

On Fri, Jan 26, 2018 at 9:28 PM, hyunkyung12 ***@***.***> wrote: easy yolo was created to reduce train time in yolo. I think the creator of easy yolo forgot to copy the data / labels folder while creating a new repository based on yolo's file. Also, if you look at the source code, you have to specify the location of the label data. Actually, there is no such directory in the easy-yolo repository. So if you create a folder called data / labels directly, the label will be displayed normally. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#9 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AMkuQaTn0M-wRqloe1GLGXlLCvisA6NQks5tOfYRgaJpZM4RViXH> .

[hyunkyung12]

in src / image.c,

image **load_alphabet()
{
    int i, j;
    const int nsize = 8;
    image **alphabets = calloc(nsize, sizeof(image));
    for(j = 0; j < nsize; ++j){
        alphabets[j] = calloc(128, sizeof(image));
        for(i = 32; i < 127; ++i){
            char buff[256];
            sprintf(buff, "data/labels/%d_%d.png", i, j);
            alphabets[j][i] = load_image_color(buff, 0, 0);
        }
    }
    return alphabets;
}

Here you can see the directory and format of the files you pasted!

[deepkshikha]

@hyunkyung12 Thanks it works for me.

[nidhinkk]

Got error while executing the make train command.

/home/username/Videos/easy-yolo-master/darknet detector train cfg/easy.data cfg/easy.cfg darknet19_448.conv.23
easy
No input parameters supplied: Success
darknet: ./src/utils.c:193: error: Assertion `0' failed.
Makefile:89: recipe for target 'train' failed
make: *** [train] Aborted (core dumped)

I am new in this training procedures.
I have some confusions in changing the easy.cfg and easy.data files. May they also leads to the error.
Can anybody help. Thank you

[hyunkyung12]

How did you enter your command?
Since all the command options are specified in the Makefile,
Just type make train to start the train.
If you want to change the command option, you can modify the Makefile.

[nidhinkk]

While configuring the model what all things needed to be calculated . Currently I am training a single class with images of resolution 640*480. Do I need to change any commands specified in the configure model part.
Thanks for the quick reply

[deepkshikha]

@hyunkyung12 Is there some way to calculate FPR value which is = FP/(FP+TN) for single class classification. Or can we draw ROC curve for one class.