Created
February 24, 2023 23:17
-
-
Save bshambaugh/c57385e5f87393677b982aa667673e9b to your computer and use it in GitHub Desktop.
axtera tech
This file contains 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
translation of: https://github.com/AXERA-TECH/ax-pipeline | |
Update Log | |
2023-01-16 Added YoloV8, please refer to ModelZoo for details. | |
2022-12-29 Added sample_v4l2_user_ivps_joint_vo for direct input of NV12 data, if the user has access to NV12 data. Modified the value of enum class SAMPLE_RUN_JOINT_MODEL_TYPE to meet the growing list of model support. | |
2022-12-26 Added sample code for rtsp input sample_rtsp_ivps_joint_vo | |
2022-12-26 Added face recognition | |
2022-12-16 Added sample for usb camera input. added license plate detection, license plate recognition, please refer to ModelZoo for details | |
2022-12-14 Added h264 file input pipeline, added yolov7-face, yolov7-palm-hand | |
2022-12-09 Added simplified pipeline build api, reduce the difficulty of building pipeline, please see new_pipeline for more details | |
2022-11-29 Added hand detection plus gesture recognition, thanks to FeiGeChuanShu, please check ModelZoo for more details. adaptive inline model input format, inline models can now not need the same input color space format | |
2022-11-24 Added Acuity Genius open source version of human detection as well as gesture model | |
2022-11-21 | |
Added pp-human-seg portrait segmentation, yolov5s-seg instance segmentation | |
New secondary inference model hrnet-pose (human pose detection based on yolov5s detection of human body after keying), configuration file MODEL_TYPE supports string and int settings, the setting value corresponds to the enumeration SAMPLE_RUN_JOINT_MODEL_TYPE or ModelZoo | |
New model details in ModelZoo | |
MODEL_PATH is added to the configuration file, user can set it, refer to hrnet_pose.json | |
2022-11-17 New yolov7-tiny, yolox-s detection model, the configuration file added the int attribute of MODEL_TYPE, this value must be set, otherwise it will not run, set the value corresponding to see the enumeration SAMPLE_RUN_JOINT_MODEL_TYPE or ModelZoo | |
2022-11-15 decoupled sample_run_joint, you can load models of different tasks at the same time, which is convenient for reasoning about multi-level model tasks like face recognition, human pose, license plate recognition, etc. | |
2022-11-14 Added adaptive NV12/RGB/BGR model, output the data format needed by the model directly through IVPS, now the model of ax-samples can be directly white whored to ax-pipeline (except yolov5 and yolov5face, other models still need to be ported to the post-processing part by the user) | |
Development boards are now supported | |
AXera-Pi (AX620A) | |
Quick start | |
Documentation | |
Quick compilation Simple cross-platform compilation based on cmake. | |
How to replace your own trained yolov5 model | |
How to deploy your own additional models | |
How to reorient images | |
ModelZoo Some supported or to be supported models, and descriptions of some models | |
Configuration file descriptions | |
Simplified version of pipeline build api | |
Examples | |
Translated with www.DeepL.com/Translator (free version) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment