convert pth to onnx

[ardeal]

Hi,

I use the following code to export pth weights to onnx. I encountered unknown error. Do you have any idea about how to export pth to onnx?

import sys
import os
import torch
from backbone import EfficientDetBackbone
def pth_onnx(input_pth):

    outpath = os.path.join(os.path.dirname(input_pth), os.path.basename(input_pth)[:-3] + 'onnx')

    compound_coef = 0
    anchor_ratios = [(1.0, 1.0), (1.4, 0.7), (0.7, 1.4)]
    anchor_scales = [2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)]

    obj_list = ['person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', 'truck', 'boat', 'traffic light',
                'fire hydrant', '', 'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep',
                'cow', 'elephant', 'bear', 'zebra', 'giraffe', '', 'backpack', 'umbrella', '', '', 'handbag', 'tie',
                'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove',
                'skateboard', 'surfboard', 'tennis racket', 'bottle', '', 'wine glass', 'cup', 'fork', 'knife', 'spoon',
                'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut',
                'cake', 'chair', 'couch', 'potted plant', 'bed', '', 'dining table', '', '', 'toilet', '', 'tv',
                'laptop', 'mouse', 'remote', 'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink',
                'refrigerator', '', 'book', 'clock', 'vase', 'scissors', 'teddy bear', 'hair drier',
                'toothbrush']

    model = EfficientDetBackbone(compound_coef=compound_coef, num_classes=len(obj_list), ratios=anchor_ratios, scales=anchor_scales)
    model.load_state_dict(torch.load(input_pth, map_location='cpu'))
    model.requires_grad_(False)
    model.eval()

    # model = EfficientDetBackbone(compound_coef=compound_coef, num_classes=len(obj_list))
    # model.load_state_dict(torch.load(f'weights/efficientdet-d{compound_coef}.pth'))
    # model.requires_grad_(False)
    # model.eval()

    dummy_input = torch.randn(1, 3, 512, 512)
    # model.set_swish(memory_efficient=False)
    with torch.no_grad():
        torch.onnx.export(model, dummy_input, outpath, opset_version=11, verbose=True)

    return

if __name__ == '__main__':

    #   ---------------------------------->
    input_pth_path = r'weights/efficientdet-d0.pth'

    pth_onnx(input_pth_path)

    aaaaa=0

[papasanimohansrinivas]

plus+1

i also have requirement to convert the model to onnx to use it on opencv

kindly help here @zylo117

it will be a huge help !