# opencv 4.12.0 — OpenCV 4.12.0

- Product: opencv (https://whatsnew.fyi/product/opencv)
- Vendor: opencv
- Date: 2025-07-02
- Version: 4.12.0
- Original notes: https://github.com/opencv/opencv/releases/tag/4.12.0
- Permalink: https://whatsnew.fyi/product/opencv/releases/4.12.0

What's New is an index, not a publisher: every entry below links to the vendor's own release notes, which are the authoritative source. Entries are labelled where they are hand-curated sample data, pre-releases, or drawn from a secondary source such as a developer blog.

Reuse: the summaries, labels and curation here are © What's New. Quote freely with attribution and a link back; wholesale republication of the corpus is not permitted — terms: https://whatsnew.fyi/terms. The vendors' own release notes remain their publishers'.

---

- **added** — Added user-defined logger callback
- **added** — Added `reinterpret()` method to `cv::Mat`
- **fixed** — Fixed empty ND-array construction
- **fixed** — Fixed bug with `int64` support for FileStorage
- **fixed** — Fixed `cv::meanStdDev` overflow for large images
- **changed** — Used HAL for cv::mean function
- **changed** — Vectorized cv::normalize / cv::norm
- **changed** — Enabled SIMD_SCALABLE for exp and sqrt
- **changed** — Vectorized normDiff with universal intrinsics
- **changed** — Refactored normDiff in hal_rvv and extended with support of more data types
- **changed** — Further vectorized copyTo with mask
- **fixed** — Fixed POWN OpenCL implementation
- **changed** — Restored legacy universal intrinsics operators in a separate header
- **changed** — Improved solveCubic accuracy
- **deprecated** — Deprecated copyData parameter in `cv::UMat` construction from `std::vector`
- **changed** — Optimized and reduced memory consumption in `cv::findContours`
- **fixed** — Fixed integer overflow in `cv::HoughLinesProbabilistic` with very large input images
- **fixed** — Fixed several potential overflows in image warping
- **changed** — Improved robustness for ellipse fitting
- **added** — Added `cv::THRESH_DRYRUN` flag to get adaptive threshold values without input thresholding
- **fixed** — Fixed `cv::getPerspectiveTransform` for singular case

*July, 2025*

Summer update for OpenCV 4.x

- Core Module:
  - Added user-defined logger callback [#27154](https://github.com/opencv/opencv/issues/27154)
  - Added `reinterpret()` method to `cv::Mat` [#25394](https://github.com/opencv/opencv/issues/25394)
  - Fixed empty ND-array construction [#27362](https://github.com/opencv/opencv/issues/27362)
  - Fixed bug with `int64` support for FileStorage [#26846](https://github.com/opencv/opencv/issues/26846)
  - Fixed `cv::meanStdDev` overflow for large images [#26867](https://github.com/opencv/opencv/issues/26867)
  - Useed HAL for cv::mean function too [#26895](https://github.com/opencv/opencv/issues/26895)
  - Vectorized cv::normalize / cv::norm [#26885](https://github.com/opencv/opencv/issues/26885)
  - Enabled SIMD_SCALABLE for exp and sqrt [#26886](https://github.com/opencv/opencv/issues/26886)
  - vectorized normDiff with universal intrinsics [#27042](https://github.com/opencv/opencv/issues/27042)
  - Refactored normDiff in hal_rvv and extended with support of more data types [#27115](https://github.com/opencv/opencv/issues/27115)
  - Further vectorized copyTo with mask [#27145](https://github.com/opencv/opencv/issues/27145)
  - Fix POWN OpenCL implementation [#27274](https://github.com/opencv/opencv/issues/27274)
  - Restored legacy universal intrinsics operators in a separate header [#27327](https://github.com/opencv/opencv/issues/27327)
  - Improved solveCubuc accuracy: [#27331](https://github.com/opencv/opencv/issues/27331), [#27347](https://github.com/opencv/opencv/issues/27347)
  - Deprecated copyData Parameter in `cv::UMat` construction from `std::vector` and always copy data [#27408](https://github.com/opencv/opencv/issues/27408)

- Imgproc Module:
  - Optimized and reduced memory consumption in `cv::findContours` [#26690](https://github.com/opencv/opencv/issues/26690), [#26834](https://github.com/opencv/opencv/issues/26834)
  - Fixed integer overflow in `cv::HoughLinesProbabilistic` with very large input images [#26744](https://github.com/opencv/opencv/issues/26744)
  - Fixed several potential overflows in image warping [#26765](https://github.com/opencv/opencv/issues/26765), [#26776](https://github.com/opencv/opencv/issues/26776)
  - Improved robustness for ellipse fitting [#26773](https://github.com/opencv/opencv/issues/26773)
  - Added `cv::THRESH_DRYRUN` flag to get adaptive threshold values without input thresholding [#26836](https://github.com/opencv/opencv/issues/26836)
  - Fixed `cv::getPerspectiveTransform` for singular case [#26926](https://github.com/opencv/opencv/issues/26926)
  - Optimized bayer2Gray with universal intrinsics [#26868](https://github.com/opencv/opencv/issues/26868)
  - Added optional mask to `cv::threshold` [#26842](https://github.com/opencv/opencv/issues/26842)
  - Accelerated `cv::remap` calculation with multi-threading, when `cv::WARP_INVERSE_MAP` is used [#27108](https://github.com/opencv/opencv/issues/27108)
  - Reworked HSV color conversion tables initialization in OpenCL branch to fix possible data raise [#27287](https://github.com/opencv/opencv/issues/27287)
  - Fixed bilateral filter's sigma color and sigma space issue [#27309](https://github.com/opencv/opencv/issues/27309)
  - Impoved performance of medianblur [#27299](https://github.com/opencv/opencv/issues/27299)
  - Added `cv::getClosestEllipsePoints` function to get the closest point on an ellipse [#26299](https://github.com/opencv/opencv/issues/26299)
  - Corrected IPP implementation for `cv::distanceTransform` results with single thread [#27432](https://github.com/opencv/opencv/issues/27432)
  - Fixed `cv::matchTemplate` with mask crash [#26703](https://github.com/opencv/opencv/issues/26703)
  - Added diamond structuring element to morphology operations [#27441](https://github.com/opencv/opencv/issues/27441), [#27492](https://github.com/opencv/opencv/issues/27492)
  - Improved `cv::intersectConvexConvex` accuracy in corner cases [#26974](https://github.com/opencv/opencv/is

_[Truncated at 4000 characters — full notes: https://github.com/opencv/opencv/releases/tag/4.12.0]_