For those who want use your own yolo model with pi5

Community Projects
1

Here I write a tutorial for you guys.

7 Likes
2

troubleshootingļ¼šCHECK failed - HEF file length does not match (status = 26)ļ¼Œ that because your host HailoData Compiler is not suitable with Pi5 HialoRT.

image
image1275Ɨ915 78.6 KB

3

And when you meet error like can not find dateset please check hailo_model_zoo/docs/DATA.rst at master Ā· hailo-ai/hailo_model_zoo Ā· GitHub to install dataset and copy dataset to the folder, if the folder is not exit please make one.

4

Some time you will find error like no alls, just do like this:

git clone https://github.com/hailo-ai/hailo_model_zoo/tree/master
cd hailo_model_zoo

Then copy hailo_model_zoo floder to the error folder

1 Like
5

I tried to use the tutorial. Instead of yolo8, i changed the lines to ā€œyolo5ā€ (I dont know if I can do this or not). then after going forward line by line according to the tutorial, when do " ```
bash run.sh object-detection-hailo


gst_parse_error: no element "hailonet" (1)
hailomuxer name=hmux filesrc location=./video/detection0.mp4 name=src_0 ! queue name=queue_dec264 max-size-buffers=3 max-size-bytes=0 max-size-time=0 !  qtdemux ! h264parse ! avdec_h264 max-threads=2 !  video/x-raw,format=I420 ! queue name=queue_scale max-size-buffers=3 max-size-bytes=0 max-size-time=0 !  videoscale n-threads=2 ! queue name=queue_src_convert max-size-buffers=3 max-size-bytes=0 max-size-time=0 !  videoconvert n-threads=3 name=src_convert
6

Can you please help me with this error?

7

Hey @jiahao.li even though some issues are being reported on the tutorial, I want to thank you very much for putting this information out there. You are helping the community greatly with this content.

2 Likes
8

Input this command and show me result

hailortcli fw-control identify

and

lspci | grep Hailo
9

narges@raspberrypi:~ $ hailortcli fw-control identify
[HailoRT] [warning] Unsupported firmware operation. Host: 4.18.0, Device: 4.17.0
Executing on device: 0000:01:00.0
Identifying board
Control Protocol Version: 2
Firmware Version: 4.17.0 (release,app,extended context switch buffer)
Logger Version: 0
Board Name: Hailo-8
Device Architecture: HAILO8L
Serial Number: HLDDLBB242602014
Part Number: HM21LB1C2LAE
Product Name: HAILO-8L AI ACC M.2 B+M KEY MODULE EXT TMP

narges@raspberrypi:~ $ lspci | grep Hailo
0000:01:00.0 Co-processor: Hailo Technologies Ltd. Hailo-8 AI Processor (rev 01)
narges@raspberrypi:~ $

10

I still get this error with hailonet.

11

image
image1035Ɨ742 70.8 KB

make sure hailort and DFC is suitableļ¼Œ and follow my tutorialļ¼š

2 Likes
12

It worked for me, but by using the provided models like yolov8n. Can you please let me know if it also works for yolov5 or not? i tried to use my yolov5 model, it didnt work.

13

The versions are now the ones its is mentioned in the image. When i used a provided model yolov8s, the codes runs successfully and detects the objects. But when i want to use my own model (yolov8n), i get this error. This is the information of both models. can you please help me with this error?:

(Hailo Detection App:11409): GStreamer-Base-CRITICAL **: 19:45:39.645: gst_queue_array_push_tail: assertion ā€˜array != NULLā€™ failed

narges@raspberrypi:~/Benchmarking-YOLOv8-on-Raspberry-PI-reComputer-r1000-and-AIkit-Hailo-8L $ hailo parse-hef ./hailomodel/yolov8n.hef
(hailo) Running command ā€˜parse-hefā€™ with ā€˜hailortcliā€™
Architecture HEF was compiled for: HAILO8L
Network group name: yolov8n, Multi Context - Number of contexts: 4
Network name: yolov8n/yolov8n
VStream infos:
Input yolov8n/input_layer1 UINT8, NHWC(640x640x3)
Output yolov8n/conv41 UINT8, FCR(80x80x64)
Output yolov8n/conv42 UINT8, NHWC(80x80x1)
Output yolov8n/conv52 UINT8, FCR(40x40x64)
Output yolov8n/conv53 UINT8, NHWC(40x40x1)
Output yolov8n/conv62 UINT8, FCR(20x20x64)
Output yolov8n/conv63 UINT8, FCR(20x20x1)

narges@raspberrypi:~/Benchmarking-YOLOv8-on-Raspberry-PI-reComputer-r1000-and-AIkit-Hailo-8L $ hailo parse-hef ./hailomodel/yolov8s_h8l.hef
(hailo) Running command ā€˜parse-hefā€™ with ā€˜hailortcliā€™
Architecture HEF was compiled for: HAILO8L
Network group name: yolov8s, Multi Context - Number of contexts: 3
Network name: yolov8s/yolov8s
VStream infos:
Input yolov8s/input_layer1 UINT8, NHWC(640x640x3)
Output yolov8s/yolov8_nms_postprocess FLOAT32, HAILO NMS(number of classes: 80, maximum bounding boxes per class: 100, maximum frame size: 160320)
Operation:
Op YOLOV8
Name: YOLOV8-Post-Process
Score threshold: 0.200
IoU threshold: 0.70
Classes: 80
Cross classes: false
Max bboxes per class: 100
Image height: 640
Image width: 640

14

Note that your outputs are not the same.
The post process in Hailo-rpi5-examples assumes you have NMS done as part of the HEF.
In order to add it you should add to your alls script a command like this:
nms_postprocess(ā€œā€¦/ā€¦/postprocess_config/yolov8n_nms_config.jsonā€, meta_arch=yolov8, engine=cpu)

You can search the alls file in our model zoo github (Not all networks are supportedā€¦)
So for examples for yolov8n see:

Note that you can use the original network output without adding the NMS, but, it requires a different post process for each network.
15

Hi @jiahao.li,

Thanks for the tutorial.
Iā€™m facing the problem stated here.
Would you be able to provide some insights as to how to work around this?

Thank you!