pub struct DetModel { /* private fields */ }Expand description
The detector session. Loaded lazily by the page worker alongside the recognizer; absent model → the pipeline runs recognition-only.
Implementations§
Source§impl DetModel
impl DetModel
Sourcepub fn load(intra: usize) -> Result<Self, String>
pub fn load(intra: usize) -> Result<Self, String>
Load the detector with intra intra-op threads (the worker’s layout
thread budget; DB is a plain conv net, so the output is stable
across thread counts to the precision a 0.3 threshold sees).
Sourcepub fn detect(&mut self, img: &RgbImage) -> Result<Vec<DetBox>, String>
pub fn detect(&mut self, img: &RgbImage) -> Result<Vec<DetBox>, String>
Detect text lines on img; boxes in img pixels, reading order.
Sourcepub fn detect_scan(
&mut self,
img: &RgbImage,
budget: bool,
) -> Result<Vec<DetBox>, String>
pub fn detect_scan( &mut self, img: &RgbImage, budget: bool, ) -> Result<Vec<DetBox>, String>
detect for a scanned page: with budget (a PDF
page render at its own 2 px/pt — not a standalone image, which
has scale 1.0 and reads at docling’s 3 px/pt effective
resolution or, over 2000 px, as decoded; not an --ocr-scale
resample), the input budgeted to the page’s own
text size (scan_budget, #945) — the full cap for small print,
down to the 960 px floor for large type — unless
DOCLING_RS_OCR_DET_BUDGET=full.
Sourcepub fn detect_capped(
&mut self,
img: &RgbImage,
max_side: u32,
) -> Result<Vec<DetBox>, String>
pub fn detect_capped( &mut self, img: &RgbImage, max_side: u32, ) -> Result<Vec<DetBox>, String>
detect under an explicit longer-side cap
(0 = none) instead of DOCLING_RS_OCR_DET_MAX_SIDE’s — the
orientation probe of a page that has a text layer (#936) reads a
few wide lines and does not keep the boxes, so it pays the
pre-#570 960 px budget, not the full one (#945).
Sourcepub fn detect_regions(
&mut self,
img: &RgbImage,
rects: &[(u32, u32, u32, u32)],
) -> Result<Vec<DetBox>, String>
pub fn detect_regions( &mut self, img: &RgbImage, rects: &[(u32, u32, u32, u32)], ) -> Result<Vec<DetBox>, String>
Detect text lines inside the given (l, t, r, b) pixel rectangles
of img — each cropped and run on its own at det_crop_size —
and return their boxes in img pixels, reading order (#945). For
the layout regions a page’s text layer leaves empty (#931): a
caption-sized crop costs a small fraction of the whole page at
the 2000 px budget and reads it at full render resolution.
Rectangles are clamped to the image; empty ones are skipped.
Auto Trait Implementations§
impl !RefUnwindSafe for DetModel
impl !UnwindSafe for DetModel
impl Forget for DetModel
impl Freeze for DetModel
impl Send for DetModel
impl Sync for DetModel
impl Unpin for DetModel
impl UnsafeUnpin for DetModel
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
impl<ST, DT> CastableFrom<ST, Initialized, Initialized> for DT
impl<ST, DT> CastableFrom<ST, Uninit, Uninit> for DT
impl<T> ErasedDestructor for Twhere
T: 'static,
Source§impl<T> IntoEither for T
impl<T> IntoEither for T
Source§fn into_either(self, into_left: bool) -> Either<Self, Self> ⓘ
fn into_either(self, into_left: bool) -> Either<Self, Self> ⓘ
self into a Left variant of Either<Self, Self>
if into_left is true.
Converts self into a Right variant of Either<Self, Self>
otherwise. Read moreSource§fn into_either_with<F>(self, into_left: F) -> Either<Self, Self> ⓘ
fn into_either_with<F>(self, into_left: F) -> Either<Self, Self> ⓘ
self into a Left variant of Either<Self, Self>
if into_left(&self) returns true.
Converts self into a Right variant of Either<Self, Self>
otherwise. Read more