|
|
@@ -180,11 +180,11 @@ class Predictor:
|
|
|
"""
|
|
|
it = iter(self.net.outputs)
|
|
|
next(it)
|
|
|
+ label_name = next(it)
|
|
|
+ label_map = np.squeeze(preds[label_name]).astype('uint8')
|
|
|
score_name = next(it)
|
|
|
score_map = np.squeeze(preds[score_name])
|
|
|
score_map = np.transpose(score_map, (1, 2, 0))
|
|
|
- label_name = next(it)
|
|
|
- label_map = np.squeeze(preds[label_name]).astype('uint8')
|
|
|
im_info = preprocessed_inputs['im_info']
|
|
|
for info in im_info[::-1]:
|
|
|
if info[0] == 'resize':
|
|
|
@@ -203,10 +203,12 @@ class Predictor:
|
|
|
def detector_postprocess(self, preds, preprocessed_inputs):
|
|
|
"""对图像检测结果做后处理
|
|
|
"""
|
|
|
- output_name = next(iter(self.net.outputs))
|
|
|
- outputs = preds[output_name][0]
|
|
|
+ outputs = self.net.outputs
|
|
|
+ for name in outpus:
|
|
|
+ if (len(outputs[name].shape == 3)):
|
|
|
+ output = preds[name][0]
|
|
|
result = []
|
|
|
- for out in outputs:
|
|
|
+ for out in output:
|
|
|
if (out[0] > 0):
|
|
|
result.append(out.tolist())
|
|
|
else:
|
syyxsxx