search

balancap / SSD-Tensorflow

Single Shot MultiBox Detector in TensorFlow
4.11k stars 1.89k forks source link

cannot draw bbox in my model #341

Open leavewave opened 5 years ago

leavewave commented 5 years ago

when i use original model, i can got a good bbox results, but with my model, i can get feedforward outputs like this, rpredictions [array([[[[[ 0.99024117, 0.00975887], [ 0.98918754, 0.01081248], [ 0.99016935, 0.00983057], [ 0.99060422, 0.0093958 ]],

     [[ 0.99471372,  0.0052863 ],
      [ 0.99353206,  0.00646789],
      [ 0.99432039,  0.0056796 ],
      [ 0.99474663,  0.00525331]],

     [[ 0.99439877,  0.00560122],
      [ 0.99402183,  0.00597819],
      [ 0.99421304,  0.00578699],
      [ 0.99402112,  0.0059789 ]],

     ..., 
     [[ 0.98937619,  0.01062382],
      [ 0.98954368,  0.01045638],
      [ 0.98972613,  0.01027387],
      [ 0.98960799,  0.01039201]],

     [[ 0.99195623,  0.00804377],
      [ 0.99115807,  0.00884192],
      [ 0.99199718,  0.00800283],
      [ 0.99241203,  0.00758795]],

but when it go through the 'np_methods.ssd_bboxes_select( rpredictions, rlocalisations, ssd_anchors, select_threshold=select_threshold, img_shape=net_shape, num_classes=21, decode=True) ' it becomes nothing, rbbox_img [ 0. 0. 1. 1.] 1111 [] 2222 [] 3333 [] 4444 [] 5555 [] sange [] [] [] no results, why, i really do not understand!