Detectron2 visualizer- 'Visualizer' object has no attribute 'get_image'

Question

I running detectron2 for object detection.

In the origina, after the trining, I run the follwing code:

v = Visualizer(im[:, :, ::-1],
            metadata=MetadataCatalog.get("name"), 
            scale=1, 
            instance_mode=ColorMode.SEGMENTATION)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"))
cv2_imshow(v.get_image()[:, :, ::-1])
cv2_imshow(resized)

and it works perfectly.

Now, after filttering some of the segmented objects, I try to run the code and present only some of the segments, so I built an array, arr_in which includes only the number of the instances that I want to present and I try to append it to the v.draw_instance_prediction.

v = Visualizer(im[:, :, ::-1],
            metadata=MetadataCatalog.get("name"), 
            scale=1, 
            instance_mode=ColorMode.SEGMENTATION   
     )
for i in range(len(arr_in)):
  num= np.int(arr_in[i])
  np.append(v, v.draw_instance_predictions(outputs["instances"].to("cpu")[num]))

cv2_imshow(v.get_image()[:, :, ::-1])
cv2_imshow(resized)

but it doesn't works. What I'm asking is how can I append variables from type predictions?

Thanks

score 1 · Answer 1 · answered Aug 03 '20 at 10:34

Solve,

v = Visualizer(im[:, :, ::-1],
            metadata=MetadataCatalog.get("name"), 
            scale=1, 
            instance_mode=ColorMode.SEGMENTATION   
)
for i in range(len(arr_in)):
  num= np.int(arr_in[i])
  #print(num)
  np.append(v.draw_instance_predictions(outputs["instances"].to("cpu")), 
     v.draw_instance_predictions(outputs["instances"].to("cpu")[num]))

v = v.draw_instance_predictions(outputs["instances"].to("cpu"))
cv2_imshow(v.get_image()[:, :, ::-1])
cv2_imshow(resized)

Detectron2 visualizer- 'Visualizer' object has no attribute 'get_image'

1 Answers1