mirror of
https://github.com/opencv/opencv.git
synced 2026-07-28 23:03:03 +04:00
Varun Jaiswal
562628eb71
Fix dnn NaN bugs in GELU SIMD and softmax for large inputs #28836 ### Bug Description - **GELU SIMD bug:** `exp(2*inner)` overflows to `inf` for large inputs (e.g. `x=10.6` → `inner≈51`), causing `inf/inf = NaN`; fixed by clamping `inner` to [-9, 9] as `tanh` already saturates to ±1.0 beyond this range. - **Softmax bug:** All `-inf` inputs (masked attention rows) produce `sum=0`, then `1/0 = inf` and `0*inf = NaN`; fixed by outputting zeros when `sum == 0`. - Add regression tests for both fixes Depends on : https://github.com/opencv/opencv/pull/28837 ### Pull Request Readiness Checklist See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request - [x] I agree to contribute to the project under Apache 2 License. - [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV - [x] The PR is proposed to the proper branch - [x] There is a reference to the original bug report and related work - [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable Patch to opencv_extra has the same branch name. - [x] The feature is well documented and sample code can be built with the project CMake