To get this error, I used trt graph created by trt.create_inference_graph(...), but if I don't create trt_graph and just use the frozen_graph after build_detection_graph(), everything runs! I assume there's something wrong with the trt.create_inference_graph(). I didn't change any argument when calling this function.
Can anyone help me? Have you had the same problem before?
I downloaded your code and used ssd_mobilenet_v1_coco for only detection task, but I get seg fault after running:
scores, boxes, classes, num_detections = tf_sess.run([tf_scores, tf_boxes, tf_classes, tf_num_detections], feed_dict={ tf_input: image_resized[None, ...] })
To get this error, I used trt graph created by trt.create_inference_graph(...), but if I don't create trt_graph and just use the frozen_graph after build_detection_graph(), everything runs! I assume there's something wrong with the trt.create_inference_graph(). I didn't change any argument when calling this function.
Can anyone help me? Have you had the same problem before?