Online Compiler: validation failed!

General
,
1

We need to compile a basic model that will help us to detect a “shampoo bottle”.

We made all the Yolo v11 training process and succesfully tested in a laptop with Cuda support (32.0.15.6624 NVIDIA GeForce RTX 4050 Laptop GPU)

Then, we tried to compile our “best.pt” file using your compiler, but with no luck…

  1. the first time we didn’t upload any custom dataset for quantization, and received the next email:

Duration: 2m8s
The compilation process failed.
Result:

{
  "error": "Failure at Hailo compiler. Please verify your model and inputs, and contact DeGirum support.",
  "input parameters": "model_file: best.pt —— model_name: botellaShampoo —— model_version: 1 —— image_height: 640 —— image_width: 640 —— cloud_zoo_url: mhbsoft/mHbZoo —— runtime_device_quant: [hailort-hailo8-quant]",
  "result": "FAILED"
}
  1. the second time we uploaded 1024 jpg images as our custom dataset, but your system gave us a RED validation message:

VALIDATION FAILED X

Could you help us to see what’s wrong?

2

Hi @jhidalgo
Please try with around 100 images. 1024 images are too many to handle for the cloud compiler. Since it is only for calibration anywhere from 64-100 images is enough. If you still encounter errors, please contact us via email (we already emailed you regarding previous failures).

3

Hi @jhidalgo, if @shashi’s response has answered your question, would you mind marking it as the solution by clicking the checkbox (:white_check_mark:) or let us know if you need more information.

Marking solutions makes it easier for others in the community who have the same question to find the answer. :slightly_smiling_face:

4

sorry, for the long delay

5

That’s quite alright @jhidalgo! Happy to see you back—thanks for marking a solution. :blush:

Feel free to create a new topic if you have any other questions!