转yolov4-tiny失败

[tosonw]

您好，我修改好了yolov4-tiny的cfg文件，但在转换yolov4-tiny为onnx时失败了，cfg文件如下： [net]

Testing

batch=1

subdivisions=1

Training

batch=64 subdivisions=1 width=416 height=416 channels=3 momentum=0.9 decay=0.0005 angle=0 saturation = 1.5 exposure = 1.5 hue=.1

learning_rate=0.00261 burn_in=1000 max_batches = 500200 policy=steps steps=400000,450000 scales=.1,.1

[convolutional] batch_normalize=1 filters=32 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=64 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky

[route] layers=-1 groups=2 group_id=1

[convolutional] batch_normalize=1 filters=32 size=3 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=32 size=3 stride=1 pad=1 activation=leaky

[route] layers = -1,-2

[convolutional] batch_normalize=1 filters=64 size=1 stride=1 pad=1 activation=leaky

[route] layers = -6,-1

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[route] layers=-1 groups=2 group_id=1

[convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky

[route] layers = -1,-2

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[route] layers = -6,-1

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[route] layers=-1 groups=2 group_id=1

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[route] layers = -1,-2

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[route] layers = -6,-1

[maxpool] size=2 stride=2

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

##################################

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=255 activation=linear

[yolo] mask = 3,4,5 anchors = 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes=80 down_stride = 32 infer_thresh = 0.25 num=6 jitter=.3 scale_x_y = 1.05 cls_normalizer=1.0 iou_normalizer=0.07 iou_loss=ciou ignore_thresh = .7 truth_thresh = 1 random=0 resize=1.5 nms_kind=greedynms beta_nms=0.6

[route] layers = -4

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[upsample] stride=2

[route] layers = -1, 23

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=255 activation=linear

[yolo] mask = 1,2,3 anchors = 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes=80 down_stride = 16 infer_thresh = 0.25 num=6 jitter=.3 scale_x_y = 1.05 cls_normalizer=1.0 iou_normalizer=0.07 iou_loss=ciou ignore_thresh = .7 truth_thresh = 1 random=0 resize=1.5 nms_kind=greedynms beta_nms=0.6

报错如下： Traceback (most recent call last): File "tools/yolo_to_onnx.py", line 748, in main() File "tools/yolo_to_onnx.py", line 743, in main verbose=True) File "tools/yolo_to_onnx.py", line 345, in build_onnx_graph conv_params) File "tools/yolo_to_onnx.py", line 219, in load_conv_weights conv_params, 'conv', 'weights') File "tools/yolo_to_onnx.py", line 248, in _create_param_tensors conv_params, param_category, suffix) File "tools/yolo_to_onnx.py", line 280, in _load_one_param_type buffer=self.weights_file.read(param_size * 4)) TypeError: buffer is too small for requested array

我也安装yolov3-tiny的样式把cfg的第二个yolo层的mask改为0，1，2，down_stride的大小也上下替换过，但都报错，请问这个是需要怎么修改？？

[LVROBOT]

@tosonw 请问yolov4-tiny 您可以使用了么？

[tosonw]

@LVROBOT 不能

[jstumpin]

To enable yolov4-tiny, only yolo_to_onnx.py needs to be patched as the existing script lacks the 'groups' support. The patch is borrowed from @jkjung-avt's repo, specifically the __make_routenode function:

`def _make_route_node(self, layer_name, layer_dict):
    """If the 'layers' parameter from the DarkNet configuration is only one index, continue
    node creation at the indicated (negative) index. Otherwise, create an ONNX Concat node
    with the route properties from the DarkNet-based graph.

    Keyword arguments:
    layer_name -- the layer's name (also the corresponding key in layer_configs)
    layer_dict -- a layer parameter dictionary (one element of layer_configs)
    """
    route_node_indexes = layer_dict['layers']
    if len(route_node_indexes) == 1:
        if 'groups' in layer_dict.keys():
            # for CSPNet-kind of architecture
            assert 'group_id' in layer_dict.keys()
            groups = layer_dict['groups']
            group_id = int(layer_dict['group_id'])
            assert group_id < groups
            index = route_node_indexes[0]
            if index > 0:
                # +1 for input node (same reason as below)
                index += 1
            route_node_specs = self._get_previous_node_specs(target_index=index)
            assert route_node_specs.channels % groups == 0
            channels = route_node_specs.channels // groups
            outputs = [layer_name + '_%d' % i for i in range(groups)]
            outputs[group_id] = layer_name
            route_node = helper.make_node(
                'Split',
                axis=1,
                split=[channels] * groups,
                inputs=[route_node_specs.name],
                outputs=outputs,
                name=layer_name,
            )
            self._nodes.append(route_node)
        else:
            index = route_node_indexes[0]
            if index > 0 :
                index +=1
            route_node_specs = self._get_previous_node_specs(target_index=index)
            layer_name = route_node_specs.name
            channels = route_node_specs.channels
    else:
        inputs = list()
        channels = 0
        for index in route_node_indexes:
            if index > 0:
                # Increment by one because we count the input as a node (DarkNet
                # does not)
                index += 1
            route_node_specs = self._get_previous_node_specs(
                target_index=index)
            inputs.append(route_node_specs.name)
            channels += route_node_specs.channels
        assert inputs
        assert channels > 0

        route_node = helper.make_node(
            'Concat',
            axis=1,
            inputs=inputs,
            outputs=[layer_name],
            name=layer_name,
        )
        self._nodes.append(route_node)
    return layer_name, channels`

yolov3-tiny-608:

yolov4-tiny-608:

[tjuskyzhang]

@tosonw 请问yolov4-tiny 您可以使用了么？

我实现了一版yolov4-tiny-tensorrt https://github.com/tjuskyzhang/yolov4-tiny-tensorrt