Introducing new API to seperate arch and board from each other#26
Merged
Conversation
Update memory_init declaration in memory_manager.h from memory_init(void) to memory_init(uintptr_t pool_start_addr). Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Change board memory backends to adopt to new memory_init. For STM32F7, define DEFAULT_MEMORY_POOL_ADDR and use it for lazy initialization call sites. For ESP32, adopt the same signature and explicitly mark the parameter unused to keep backend behavior unchanged. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Replace single-platform selection with explicit EMBEDDIP_TARGET_BOARD, EMBEDDIP_ARCH and EMBEDDIP_CPU cache variables and enforce a strict compatibility matrix at configure time. Load board and arch profiles independently, then compose sources, include dirs, defines and compile options from those profiles so new ports can be added without touching central build logic. Move package config template to embedDIP root and update configure_package_config_file path accordingly. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Remove board level common/fft implementation units and keep board ownership limited to board specific memory/config responsibilities. Update board memory implementation to match the new split and add the new ESP32 board memory source naming/layout used by the refactor. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Move device implementations out of nested vendor subfolders into the top level device category paths and update build references to the new locations. This reduces path depth, makes board profile source lists simpler and keeps driver discovery consistent across STM32 and ESP32 backends. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Introduce a shared imgproc FFT implementation for common 2D flow and move arch-specific math calls behind a small backend hook interface. Implement ARM and XTENSA backend hooks in arch specific files and remove duplications. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Update embedDIP public headers/config checks to validate one board and one arch/cpu selection under the new model. Adopt EMBED_DIP_BOARD_* and EMBED_DIP_ARCH_* style compile contracts and drop legacy TARGET_BOARD_* assumptions from the public interface. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
ozan956
force-pushed
the
feature/new-api
branch
from
ec94922 to
7746603
Compare
Update EmbedDIP semantic version from 0.1.0 to 0.2.0 to reflect build/config contract changes introduced by the board/arch/cpu decoupling and related refactors. This release marks a compatibility boundary for consumers migrating from the legacy single-platform selection flow. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
ozan956
force-pushed
the
feature/new-api
branch
from
daabea3 to
9d593a0
Compare
Drop the C++ wrapper declaration/definition for Image::ifftshift(). The wrapper only forwarded to fftshift() and provided no distinct behavior. Keeping a single shift API simplifies ImageWrapper surface area and avoids duplicate frequency-shift entry points. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Introduce board and architecture profile files for STM32F7/ESP32 and ARM/XTENSA targets, and move source/define/include selection into those profiles to support board-arch-cpu split configuration. Adjust .gitignore CMake patterns so new profile cmake files are tracked. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Drop large commented-out legacy UART implementations from stm32_uart.c to keep the active serial backend clear and maintainable. Also remove the per-chunk HAL_Delay(1) in serial_send_jpeg(), relying on normal blocking UART transmit behavior for chunk pacing. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Introduce a new imgproc compression module. Implement JPEG encoding via libjpeg when available, including a fixed-capacity destination manager with overflow detection and input support for various images. When libjpeg is not present, keep a safe stub that returns -1. Wire the module into the build by adding sources, STM32F7 libjpeg include paths, and EMBEDDIP_HAVE_LIBJPEG compile definition when jpeglib.h is found. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Drop unfinished/duplicate GrabCut entry points and consolidate the API around grabCut() and grabCutLite(). Implement a memory-aware grayscale graph-cut path with: - ROI-bounded processing and argument/format validation - adaptive downsampling for embedded node limits - Dinic-style maxflow (BFS/DFS residual graph) - hard border/background constraints and n-link smoothness terms - mask upsampling back to full-resolution ROI Update public headers and C++ wrapper accordingly: - add Image::grabCut(mask, roi, iterations) returning status - remove obsolete RGB/legacy wrapper methods Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Fix several robustness bugs in NMS/double-threshold/hysteresis and tighten Canny error handling. This removes crash-prone paths and makes edge linking behavior deterministic on embedded targets. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Use one global visited map and one global stack across all seeds. Switch grayscale and color growth to adaptive running region mean grow by distance to region mean instead of fixed seed value. Return INVALID_ARG when no valid seeds are provided. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Allow memory_init() to accept either an SDRAM offset or an absolute SDRAM address for compatibility, validate computed bounds, and fall back to the default reserved pool start when input is invalid or leaves no room for allocator metadata. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Add clarifying comments in erode() and dilate() that the ping buffer is initialized from source pixels before the loop so the first iteration runs on original image data. Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
Signed-off-by: Ozan Durgut <ozandurgut.2001@hotmail.com>
This was
linked to
issues
Apr 23, 2026
This was
unlinked from
issues
Apr 23, 2026
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
1 participant
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.
No description provided.