Dear Community,
I am new to Hailo (and AI in general), exploring the Hailo8L on RPi 5. I’ve successfully installed and run the basic pipelines in the hailo-rpi5-examples repo. Now similar to pose estimation, I am trying to implement hand landmark detection.
I’ve found a compiled hand_landmark_lite.hef file for Hailo8L in the model zoo. However, I’m unsure about the appropriate post-processing file to use. Specifically, I need to modify this line:
“self.default_postprocess_so = os.path.join(self.postprocess_dir, ‘libyolov8pose_post.so’)”
In my /usr/lib/aarch64-linux-gnu/hailo/tappas/post-process folder I find the following .so files, which includes facial landmarks but not for hand landmarks:
libocr_post.so, libcenterpose_post.so, libperson_attributes_post.so, libclassification.so, libre_id.so, libdebug.so, libscrfd_post.so, libdepth_estimation.so, libsemantic_segmentation.so, libface_attributes_post.so, libstream_id_tool.so, libface_detection_post.so, libyolo_hailortpp_post.so, libface_recognition_post.so, libyolo_post.so, libfacial_landmarks_post.so, libyolov5seg_post.so, libmobilenet_ssd_post.so, libyolov8pose_post.so, libmspn_post.so, libnanodet_post.so
My questions are:
- Is there an existing post-processing file I can use for hand landmark detection?
- If not, could you guide me through the steps to create this file using the existing .hef file?
Any help or guidance would be greatly appreciated. Thanks!