[Issue]: Inconsistent Python API

Open
#181 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
35/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Quiet
Tech stack
cpp, python

Research direction

Start by comparing the rocCV Python bindings and pbdocs for the listed operators with the referenced CVCUDA operator definitions. Audit naming, optional defaults, extra parameters, and into/non-into signatures across the full list. Done means the rocCV Python API and documentation consistently match the intended CVCUDA-compatible interface.

Written by the indexing model from the issue text.

Description

There are some inconsistencies between the rocCV and CVCUDA pybindings / Python API which make it harder for rocCV to be a true "drop-in". Both libraries are inconsistent in their naming (snake vs camel case), and there are also some places where CVCUDA marks params as optional with default values, while rocCV has them required.

CVCUDA code reference used: Github.

  • AdvCvtColor
    • conversion_code vs code
    • color_spec vs spec
  • BilateralFilter
    • sigmaColor vs sigma_color
    • sigmaSpace vs sigma_space
    • borderMode vs border
      • Note that the rocCV pbdoc describes it as border_mode
      • border is optional with default value in CVCUDA
    • rocCV takes an extra borderValue parameter
      • Note that the rocCV pbdoc describes it as border_value
  • BndBox
    • bnd_boxes vs bboxes
  • CenterCrop
    • rocCV pbdocs errenously has a dst param for the non-into function
  • ConvertTo
    • Op name: convert_to vs convertto
    • alpha vs scale
    • beta vs offset
  • CustomCrop
    • Op name: custom_crop vs customcrop
    • crop_rect vs rect
    • rocCV pbdocs errenously has a dst param for the non-into function, while missing that param for the into function
  • CvtColor
    • conversion_code vs code
  • Flip
    • flip_code vs flipCode
  • Gaussian
    • kernelSize vs kernel_size
    • borderMode vs border
  • Reformat
    • input vs src
    • out_layout vs layout
  • Remap
    • in_interpolation vs src_interp
      • src_interp is optional with default value in CVCUDA
    • map_interpolation vs map_interp
      • map_interp is optional with default value in CVCUDA
    • map_value_type vs map_type
      • map _type is optional with default value in CVCUDA
    • border_type vs border
      • border is optional with default value in CVCUDA
    • align_corners is optional with default value in CVCUDA
    • border_value is optional with default value in CVCUDA
  • Resize
    • interp is optional with default value in CVCUDA
  • Thresholding
    • maxVal vs maxval
    • threshType vs type
    • rocCV has extra maxBatchSize param
  • WarpAffine
    • CVCUDA accepts a flags param to control the interpolation method and whether it is inverse, while rocCV accepts interp and inverted params
    • border_mode is optional with default value in CVCUDA
    • border_value is optional with default value in CVCUDA
  • WarpPerspective
    • CVCUDA accepts a flags param to control the interpolation method and whether it is inverse, while rocCV accepts interp and inverted params
Dominant language
C++
Stars
1
Forks
10
Avg merge
3d 1h
Merged PRs (30d)
1

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from ROCm/rocCV

All issues in ROCm/rocCV

Similar issues

More C++ issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.