Closed ternaus closed 4 months ago
This pull request adds support for handling negative values in addition and multiplication operations for uint8 and float32 images. It includes new test cases to ensure the correctness of these operations and updates benchmark results to reflect the performance of the new implementation. Significant changes include modifying the 'prepare_value_opencv' and 'apply_numpy' functions to handle negative values correctly, and updating the 'convert_value' utility function to simplify its logic.
Files | Changes |
---|---|
benchmark/results/float32_1.md benchmark/results/float32_3.md benchmark/results/float32_5.md benchmark/results/uint8_1.md benchmark/results/uint8_3.md benchmark/results/uint8_5.md |
Updated benchmark results to include new performance metrics for albucore 0.0.10. |
All modified and coverable lines are covered by tests :white_check_mark:
Please upload report for BASE (
main@0a706fe
). Learn more about missing BASE report.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Summary by Sourcery
This pull request enhances the
add
,multiply
,add_weighted
, andmultiply_add
functions to handle negative values and weights. It also includes extensive new test cases to ensure the correctness of these enhancements and updates the benchmark results to reflect the performance changes.add
function to handle negative values and added corresponding test cases.multiply
function to handle negative values and added corresponding test cases.add_weighted
function to handle negative weights and added corresponding test cases.multiply_add
function to handle negative factors and values, with new test cases.power
function to handle various exponent formats and added new test cases.add
,multiply
,add_weighted
, andmultiply_add
functions.