Hi,
I have been trying to run custom ONXX model on eagle ai studio for object detection. Below steps I followed-
Model analyzer -> selected AM69A - 32 TOPS -> selected custom model onnx runtime
Then I downloaded the pre trained model from this link https://github.com/onnx/models/tree/main/vision/object_detection_segmentation/mask-rcnn (Mask R-CNN R-50-FPN) and tried to run.
Each time I am trying to compile the model, the kernel is getting dead and in some cases, it is giving bad malloc error. This happens when I am using custom onnx model only. So far I am not able to compile and run any single onnx custom model. I have 3 models.
Mask RCNN, yolact (resnet50) and yolact (resnet18).
Even I tried the simple onxx code to check the model -
import onnx import os # Preprocessing: load the ONNX model #model_path = os.path.join("resources", "single_relu.onnx") onnx_model = onnx.load('path to uploaded model') print("The model is:\n{}".format(onnx_model)) # Check the model try: onnx.checker.check_model(onnx_model) except onnx.checker.ValidationError as e: print("The model is invalid: %s" % e) else: print("The model is valid!")
Here also the kernel keeps getting disconnected and dead. I tried 1 more custom model onnx based but got the same issue. I have to do log out and login to make it work again. Restarting EVM or kernel is not fixing this issue.
Please let me know why is it happening.
Thanks