> For the complete documentation index, see [llms.txt](https://ncxlib.gitbook.io/ncxlib/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ncxlib.gitbook.io/ncxlib/getting-started/api-documentation/overview/preprocessor/imagegrayscaler.md).

# ImageGrayscaler

```python
class ImageGrayscaler(Preprocessor):
    def __init__(self):
        super().__init__()

    def img_to_grayscale(self, image: np.ndarray) -> np.ndarray:
        grayscale_image = []
        for r, g, b in image:
            gray = int(0.299 * r + 0.587 * g + 0.114 * b)
            grayscale_image.append(gray)

        return np.array(grayscale_image)

    def convert_all_img_to_grayscale(self, dataset: Dataset) -> pd.DataFrame:
        imgs = []

        data = dataset.data.copy()
        for _, row in data.iterrows():
            rgb_pixels = row["data"]
            grayscale_image = self.img_to_grayscale(rgb_pixels)
            img_array = np.array(grayscale_image)
            imgs.append({"title": row["title"], "data": img_array, "target": row["target"]})

        dataframe = pd.DataFrame(imgs)
        dataframe["title"] = dataframe["title"].astype("string")
        if not dataset.label_numeric:
            dataframe["target"] = dataframe["target"].astype("string")

        return dataframe

    def apply(self, dataset: Dataset) -> Dataset:
        dataset.data = self.convert_all_img_to_grayscale(dataset)
        return dataset
```