|
|
@@ -365,8 +365,8 @@ darknet.Graph = class {
|
|
|
layer.out_w = params.w;
|
|
|
layer.out_c = params.c;
|
|
|
layer.out = layer.in;
|
|
|
- load_batch_normalize_weights(weights, section, '', layer.out);
|
|
|
- layer.outputs[0].type = new darknet.TensorType('float32', make_shape([ layer.ouputs ], 'batchnorm'));
|
|
|
+ load_batch_normalize_weights(layer, '', layer.out_c);
|
|
|
+ layer.outputs[0].type = new darknet.TensorType('float32', make_shape([ layer.out_w, layer.out_h, layer.out_c ], 'batchnorm'));
|
|
|
break;
|
|
|
}
|
|
|
case 'activation': {
|
|
|
@@ -374,7 +374,7 @@ darknet.Graph = class {
|
|
|
layer.out_w = params.w;
|
|
|
layer.out_c = params.c;
|
|
|
layer.out = layer.in;
|
|
|
- layer.outputs[0].type = new darknet.TensorType('float32', make_shape([ layer.ouputs ], 'activation'));
|
|
|
+ layer.outputs[0].type = new darknet.TensorType('float32', make_shape([ layer.out_w, layer.out_h, layer.out_c ], 'activation'));
|
|
|
break;
|
|
|
}
|
|
|
case 'max':
|