vitali-fedulov / images4 Goto Github PK

Hi, it's an awesome project for duplicate image removal. I use this project to detect and delete images, but I encountered some issues.

For example, there are two thumbnails which are very alike from my raw eyes. But images4.Similar returns false in this case.

Is there any way to optimize this situation?

Could you please add the MarshalJSON function to null the Icon type in JSON encoding, or add the json:",omitempty" flag to omit empty values in Icon and ImgSize types?

I have loads of images in my data and when I want to browse through them in JSON format, I get bigger than necessary JSON files since there are lots of empty images. 🙃

Current marshal result:

{
    "Icon": {
        "Pixels": null,
        "ImgSize": {
            "X": 0,
            "Y": 0
        }
    }
}

You can imagine how big it gets when there are lots of empty icons in the array. 😵

Suggestion 1 - MarshalJSON function

This results in smaller size than the suggestion 2.

func (i IconT) MarshalJSON() ([]byte, error) {
    if i.Pixels == nil {
        return []byte("null"), nil
    }
    
    return json.Marshal(struct {
        Pixels  []uint16
        ImgSize Point
    }{i.Pixels, i.ImgSize})
}

Output:

{
    "Icon": null
}

Suggestion 2 - json:",omitempty" flag

This reduces the size by removing the "Pixels", "X" and "Y" fields.

type IconT struct {
    Pixels    []uint16 `json:",omitempty"` // Visual signature.
    ImgSize image.Point `json:",omitempty"` // Original image size.
}

type Point struct {
    X, Y int `json:",omitempty"`
}

Output:

{
    "Icon": {
         "ImgSize": {}
    }
}

Suggestion 3

Do you have any better suggestions? 🤔