iVSR model JSON config and enablement#41
Open
sandeshk-intel wants to merge 7 commits into
Open
Conversation
Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
… path - Fix Y-plane row-stride bug in unpack_output_window: use output->width instead of unpadded frame width 'fw' as the tensor stride. Fixes corrupted output on frames not a multiple of 'align' (e.g., 720p, 854p). - Flip normalize_input/output defaults from 1 to 0. Most models use raw [0,255] I/O; old defaults silently corrupted VideoSeal/EDSR/VideoProc. - Fix model_color fallback to default to "RGB" instead of an empty string when absent, preventing SDK undefined behavior without heap allocation. - Reorder color-format dispatch so color_format_auto is evaluated before model_color, preventing VideoProc/CustVSR auto-detection shadowing. - Inherit out_layout from in_layout when absent from the JSON config. Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
Update documentation, fix the VideoProc normalize_output value.
Signed-off-by: sandeshk <sandesh.kumar.s@intel.com>
sandeshk-intel
requested review from
linxie47,
wangjingz and
xiaoxial
as code owners
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Introduces a JSON-based model configuration system for the iVSR DNN backend, replacing hardcoded per-model C descriptors with external config files.
Add JSON config files for all supported models (VideoProc, EDSR, CustVSR, TSENet, RIFE, VideoSeal, HDRTVNet-LE) and a README documenting all parameters
Fix Y-plane row-stride bug in unpack_output_window that corrupted output on non-aligned frame widths (720p, 854p)
Flip normalize_input/normalize_output defaults from 1 to 0 — old defaults silently corrupted VideoSeal, EDSR, and VideoProc output
Fix model_color fallback to "RGB" instead of empty string; reorder color-format dispatch so color_format_auto takes precedence over model_color
Inherit out_layout from in_layout when absent from config
Remove redundant chroma copy for SVP/CustVSR models
Set out_precision_depth_derived=false for HDRTVNet-LE and fix normalize_output value in VideoProc config