rlx-ocr 0.2.0

OCR engine for RLX — text detection + recognition
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204

// RLX — versatile ML compiler + runtime.
// Copyright (C) 2026 Eugene Hauptmann, Nataliya Kosmyna.
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, version 3.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.

//! Greyscale image preprocessing for OCR models.

use rten_tensor::prelude::*;
use rten_tensor::{NdTensor, NdTensorView};
use std::fmt;

/// Normalized greyscale background value used by ocrs models (matches `ocrs` 0.12.x).
pub const BLACK_VALUE: f32 = -0.5;

/// Errors when constructing an [`ImageSource`].
#[derive(Clone, Debug, PartialEq, Eq)]
pub enum ImageSourceError {
    UnsupportedChannelCount,
    InvalidDataLength,
}

impl fmt::Display for ImageSourceError {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        match self {
            Self::UnsupportedChannelCount => f.write_str("channel count is not 1, 3 or 4"),
            Self::InvalidDataLength => {
                f.write_str("data length is not a multiple of width * height")
            }
        }
    }
}

impl std::error::Error for ImageSourceError {}

/// Pixel layout for image tensors.
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
pub enum DimOrder {
    Hwc,
    Chw,
}

enum ImagePixels<'a> {
    #[allow(dead_code)]
    Floats(NdTensorView<'a, f32, 3>),
    Bytes(NdTensorView<'a, u8, 3>),
    FloatsOwned(NdTensor<f32, 3>),
}

/// Input image for [`crate::OcrEngine::prepare_input`].
pub struct ImageSource<'a> {
    data: ImagePixels<'a>,
    order: DimOrder,
}

impl<'a> ImageSource<'a> {
    /// RGB/RGBA/greyscale bytes in HWC order (`image` crate layout).
    pub fn from_bytes(bytes: &'a [u8], dimensions: (u32, u32)) -> Result<Self, ImageSourceError> {
        let (width, height) = dimensions;
        let channel_len = (width as usize).saturating_mul(height as usize);
        if channel_len == 0 {
            return Err(ImageSourceError::UnsupportedChannelCount);
        }
        if !bytes.len().is_multiple_of(channel_len) {
            return Err(ImageSourceError::InvalidDataLength);
        }
        let chans = bytes.len() / channel_len;
        if !matches!(chans, 1 | 3 | 4) {
            return Err(ImageSourceError::UnsupportedChannelCount);
        }
        let view = NdTensorView::from_data([height as usize, width as usize, chans], bytes);
        Ok(Self {
            data: ImagePixels::Bytes(view),
            order: DimOrder::Hwc,
        })
    }

    /// Existing CHW or HWC float tensor (copied).
    pub fn from_tensor(
        tensor: NdTensorView<'_, f32, 3>,
        order: DimOrder,
    ) -> Result<ImageSource<'static>, ImageSourceError> {
        let chans = match order {
            DimOrder::Hwc => tensor.size(2),
            DimOrder::Chw => tensor.size(0),
        };
        if chans == 0 || !matches!(chans, 1 | 3 | 4) {
            return Err(ImageSourceError::UnsupportedChannelCount);
        }
        let owned = NdTensor::from_data(tensor.shape(), tensor.to_vec());
        Ok(ImageSource {
            data: ImagePixels::FloatsOwned(owned),
            order,
        })
    }
}

/// Convert an image to a normalized greyscale CHW tensor `[1, H, W]`.
pub fn prepare_image(img: ImageSource<'_>) -> NdTensor<f32, 3> {
    match (&img.data, img.order) {
        (ImagePixels::Floats(f), DimOrder::Hwc) => prepare_floats::<true>(f.view()),
        (ImagePixels::Floats(f), DimOrder::Chw) => prepare_floats::<false>(f.view()),
        (ImagePixels::FloatsOwned(f), DimOrder::Hwc) => prepare_floats::<true>(f.view()),
        (ImagePixels::FloatsOwned(f), DimOrder::Chw) => prepare_floats::<false>(f.view()),
        (ImagePixels::Bytes(b), DimOrder::Hwc) => prepare_bytes::<true>(b.view()),
        (ImagePixels::Bytes(b), DimOrder::Chw) => prepare_bytes::<false>(b.view()),
    }
}

fn prepare_floats<const CHANS_LAST: bool>(floats: NdTensorView<'_, f32, 3>) -> NdTensor<f32, 3> {
    const ITU: [f32; 3] = [0.299, 0.587, 0.114];
    let n = if CHANS_LAST {
        floats.shape()[2]
    } else {
        floats.shape()[0]
    };
    match n {
        1 => convert_pixels::<f32, 1, 1, CHANS_LAST>(floats, [1.]),
        3 => convert_pixels::<f32, 3, 3, CHANS_LAST>(floats, ITU),
        4 => convert_pixels::<f32, 4, 3, CHANS_LAST>(floats, ITU),
        _ => panic!("expected greyscale, RGB or RGBA input image"),
    }
}

fn prepare_bytes<const CHANS_LAST: bool>(bytes: NdTensorView<'_, u8, 3>) -> NdTensor<f32, 3> {
    const ITU: [f32; 3] = [0.299, 0.587, 0.114];
    let weights = ITU.map(|w| w / 255.0);
    let n = if CHANS_LAST {
        bytes.shape()[2]
    } else {
        bytes.shape()[0]
    };
    match n {
        1 => convert_pixels::<u8, 1, 1, CHANS_LAST>(bytes, [1. / 255.0]),
        3 => convert_pixels::<u8, 3, 3, CHANS_LAST>(bytes, weights),
        4 => convert_pixels::<u8, 4, 3, CHANS_LAST>(bytes, weights),
        _ => panic!("expected greyscale, RGB or RGBA input image"),
    }
}

fn convert_pixels<
    T: Copy + Into<f32>,
    const PIXEL_STRIDE: usize,
    const CHANS: usize,
    const CHANS_LAST: bool,
>(
    src: NdTensorView<'_, T, 3>,
    chan_weights: [f32; CHANS],
) -> NdTensor<f32, 3> {
    let [height, width, chans] = if CHANS_LAST {
        src.shape()
    } else {
        let [c, h, w] = src.shape();
        [h, w, c]
    };
    assert_eq!(chans, PIXEL_STRIDE);

    let mut out_pixels = Vec::with_capacity(height * width);
    if CHANS_LAST {
        let src = src.to_contiguous();
        let mut iter = src.data().chunks_exact(PIXEL_STRIDE);
        debug_assert!(iter.remainder().is_empty());
        for in_pixel in iter.by_ref() {
            let mut pixel = BLACK_VALUE;
            for (c, &w) in chan_weights.iter().enumerate() {
                pixel += in_pixel[c].into() * w;
            }
            out_pixels.push(pixel);
        }
    } else {
        for y in 0..height {
            out_pixels.extend((0..width).map(|x| {
                let mut pixel = BLACK_VALUE;
                for (c, &w) in chan_weights.iter().enumerate() {
                    pixel += src[[c, y, x]].into() * w;
                }
                pixel
            }));
        }
    }
    NdTensor::from_data([1, height, width], out_pixels)
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn preprocess_greyscale_bytes() {
        let out = prepare_image(ImageSource::from_bytes(&[0, 128, 255, 64], (2, 2)).unwrap());
        assert_eq!(out.shape(), [1, 2, 2]);
        assert!((out[[0, 0, 0]] - BLACK_VALUE).abs() < 1e-5);
        assert!((out[[0, 0, 1]] - (BLACK_VALUE + 128.0 / 255.0)).abs() < 1e-5);
    }
}