yunyang1994 / yymnist Goto Github PK

=== before ===

for i in range(w):
      for j in range(h):
            x = xmin + i
            y = ymin + j
            blank[y][x] = image[j][i]

=== after ===
blank[ymin:ymin+h,xmin:xmin+w] = image[0:h,0:w]

return ((xmax-xmin) * (ymax - ymin)) / (A1 + A2)

in this line, i think it should return ((xmax-xmin) * (ymax - ymin)) / (A1 + A2 - (xmax-xmin) * (ymax - ymin)) according to iou's definition.

如果能增加跟导出的图片数量相同的labels文件，将可以使用yolo进行训练。

选择相应路径，运行make_data.py后，正确读取了./mnist/train/中的数据，也创建了Images文件夹和labels.txt，但是Images文件夹中并没有.jpg的图片，labels.txt中却有了imags路径以及标签，寻求您解答为什么Images文件夹中写入图片。

image is read randomly in each loop. this mean that some image from the dataset might be missed, but some might be read a few times.

idx = random.randint(0, total_image)
data[0] = make_image(data, image_paths[idx], ratio)

https://github.com/YunYang1994/yymnist#intorduction
Could you please change from "intorduction" to "introduction"