Memory size frame count mismatch

General
1

Hello there

I recently received the Raspberry Pi PCIe Accelerator Module with the Hailo-8L and was able to run the examples from the rpi5-examples and the Hailo Application Code Examples (GitHub - hailo-ai/Hailo-Application-Code-Examples). However, suddenly I get the an error when running any of the examples from Hailo Application Code Examples. This specific error happens when running the object detection example. I downloaded the correct hef for the Hailo-8L from the model zoo repository.

[HailoRT] [error] CHECK failed - Memory size of vstream yolov7/input_layer1 does not match the frame count! (Expected 1228800, got 0)

I can successfully run the hef using hailort-cli run. Does anybody have any pointers to what I am doing wrong?

2

Hey @funs

Welcome to Hailo Community.

is this the python or CPP examples ?
Please add the exact application you are running and make sure that the repo is updated , if you are using python on the rpi5 ( hailo_platform) it is not yet supported , will be enabled soon.

Regards

3

Hello Omri

Thank you for your response. Yes, I was indeed trying to run the python examples. I was able to build and run the C++ examples, which are working fine.

I look forward to Python support on the Pi 5 and will be using the C++ API for my stuff.

Thanks, kind regards
Stefan

1 Like
4

I am also facing the same problem. Earlier the example code from GitHub - hailo-ai/Hailo-Application-Code-Examples was working for object-detection python but now suddenly giving the same error, I am also using Hailo-8L

[HailoRT] [error] CHECK failed - Memory size of vstream ssd_mobilenet_v1/input_layer1 does not match the frame count! (Expected 270000, got 0)
[HailoRT] [error] CHECK_SUCCESS failed with status=HAILO_INVALID_ARGUMENT(2)

can any one help?

1 Like
5

Hi @jamesbonr123,
Which hailort version are you using?

6

Hi @nina-vilela
I am using v4.18.0 version of hailort

7

Hi @jamesbonr123,

We believe that the issue was fixed. Could you give it a try with the updated code?

8

I have this error too. I work on raspberry Pi 5 and installed hailort, pcie driver and python API package for python 3.11 with version 4.18 manuallly.
On raspberry both hailortcli benchmark ~/yolov8n.hef and running the model from python API works (on tracker hailo code example).

The problem occurs when I run the code inside docker container (I installed hailort==4.18 and python API package for python 3.10 there).
user@docker:~$ hailortcli benchmark hailo/yolov8n.hef works but running the model through the python API does not. Is there anything I can do to fix this?

9

The problem was that I used incorrect version of numpy (above > 2.0) whereas it is required to run it with numpy==1.23.3

5 Likes
10

Thanks for letting us know the solution. We will consider adding a numpy version requirement.

11

thank you for providing solution!

I’m also strugging the same problm, but after I reinstalled numpy==1.23.3, I managed to solve this!

12

I set numpy to version 1.23.3 but I’ve the same issue.

13

I had the same issue with numpy 1.26.4 on my raspberry pi 5
Control Protocol Version: 2
Firmware Version: 4.20.0 (release,app,extended context switch buffer)
Logger Version: 0
Board Name: Hailo-8

the model scdepthv3.hef expected a batch size with sending the frames:

    # adding batch dimension: (H, W, C) -> (1, H, W, C) for NHWC format

    frame = np.expand_dims(frame, axis=0)
14

Hey @Ralph_Muthreich,

Welcome to the Hailo Community!

Could you try using numpy version 1.23.3? That should help for now. This issue will get fixed once the trixiy package is updated to work with numpy 2!