--input_shape

Description

Sets the shape of each input.

See Also

None

Arguments

Argument:

Shape information of the model input, for example, "input_name1:n1,c1,h1,w1;input_name2:n2,c2,h2,w2". Enclose the whole argument in double quotation marks (""), and separate the nodes by semicolons (;). input_name must be the node name in the network model before model conversion.

Restrictions:

If a dimension of the input data for the original model is not fixed, for example, input_name1:?,h,w,c, this option is mandatory. The question mark (?) indicates the batch size (number of images processed at a time). You can perform the following operations:

1. Set ? to a fixed value (such as 1, 2, 3...) to convert the original model with dynamic shape into a model with static shape.
2. If the input data of a model is a scalar (of type bool, for example), you do not need to set --input_shape of the node.
3. Set ? to –1 and use this option in conjunction with --dynamic_batch_size to set the dynamic batch size. For details, see --dynamic_batch_size.

Suggestions and Benefits

None

Examples

--input_shape="input_name1:n1,c1,h1,w1;input_name2:n2,c2,h2,w2"

Dependencies and Restrictions

None