ImgAug_yolov5

This approach uses Data Augmentation to generate new samples given a training/validation dataset without the Keras Augmentation.

Prerequisites They are the same as YOLOv5, but make sure you have already installed them.

Recall: YOLOv5 requires the dataset to be in the darknet format. Here’s an outline of what it looks like:

• One txt with labels file per image
• One row per object
• Each row is class x_center y_center width height format.
• Box coordinates must be in normalized xywh format (from 0 - 1). If your boxes are in pixels, divide x_center and width by image width, and y_center and height by image height.

Class numbers are zero-indexed (start from 0). Example:

Image properties: width=1156 pix, height=1144 pix.
bounding box properties: xmin=1032, ymin=20, xmax=1122, ymax=54, object_name="Ring".
Let objects_list="bracelet","Earring","Ring","Necklace"
YOLOv5 format: f"{category_idx} {x1 + bbox_width / 2} {y1 + bbox_height / 2} {bbox_width} {bbox_height}\n"

𝑏𝑏𝑜𝑥𝑤𝑖𝑑𝑡ℎ=𝑥𝑚𝑎𝑥/𝑤𝑖𝑑𝑡ℎ−𝑥𝑚𝑖𝑛/𝑤𝑖𝑑𝑡ℎ=(1122−1032)/1156=0.07785467128027679 
𝑏𝑏𝑜𝑥ℎ𝑒𝑖𝑔ℎ𝑡=𝑦𝑚𝑎𝑥/ℎ𝑒𝑖𝑔ℎ𝑡−𝑦𝑚𝑖𝑛/ℎ𝑒𝑖𝑔ℎ𝑡=(54−20)/1144=0.029720279720279717 
𝑥𝑐𝑒𝑛𝑡𝑒𝑟=𝑥𝑚𝑖𝑛/𝑤𝑖𝑑𝑡ℎ+𝑏𝑏𝑜𝑥𝑤𝑖𝑑𝑡ℎ/2=0.9316608996539792 
𝑦𝑐𝑒𝑛𝑡𝑒𝑟=𝑦𝑚𝑖𝑛/ℎ𝑒𝑖𝑔ℎ𝑡+𝑏𝑏𝑜𝑥ℎ𝑒𝑖𝑔ℎ𝑡/2=0.032342657342657344 
category_idx=2
Final result: 2 0.9316608996539792 0.032342657342657344 0.07785467128027679 0.029720279720279717

some functions:

• Data Reading and Storage Functions

• Photometric Transformations

• Geometric Transformations¶

• Random Occlusion

• Deep Learning based Approaches (experimental)

    Suggested Labeling for TTA
    gaussian noise: _GN
    localvar noise: _LN
    poisson noise: _PN
    salt noise: _SN
    pepper noise: _PP
    salt&pepper: _SP
    speckle noise:_SE
    gray: _GR
    Histogram Equalization: _HE
    shear x: _SX
    shear y: _SY
    flip lr: _LR
    flip ud: _UD
    rotation 90: _R90
    rotation 180: _R180
    rotation 270: _R270
    random erasing: img _RE