Why pytorch‘s effect is so bad?

[As-Time-Goes-By]

hi after running test，the result occurs. [2023-03-11 10:30:10,134 INFO test.py line 299 38140] Class_0 Result: iou/accuracy 0.0014/0.0015, name: ceiling. [2023-03-11 10:30:10,134 INFO test.py line 299 38140] Class_1 Result: iou/accuracy 0.1666/0.5595, name: floor. [2023-03-11 10:30:10,134 INFO test.py line 299 38140] Class_2 Result: iou/accuracy 0.0557/0.0652, name: wall. [2023-03-11 10:30:10,134 INFO test.py line 299 38140] Class_3 Result: iou/accuracy 0.0000/0.0001, name: beam. [2023-03-11 10:30:10,134 INFO test.py line 299 38140] Class_4 Result: iou/accuracy 0.0112/0.0868, name: column. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_5 Result: iou/accuracy 0.0021/0.0022, name: window. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_6 Result: iou/accuracy 0.0224/0.0665, name: door. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_7 Result: iou/accuracy 0.0042/0.0064, name: chair. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_8 Result: iou/accuracy 0.0063/0.0262, name: table. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_9 Result: iou/accuracy 0.0000/0.0000, name: bookcase. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_10 Result: iou/accuracy 0.0019/0.0019, name: sofa. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_11 Result: iou/accuracy 0.0056/0.0749, name: board. [2023-03-11 10:30:10,135 INFO test.py line 299 38140] Class_12 Result: iou/accuracy 0.0077/0.0083, name: clutter. [2023-03-11 10:30:10,135 INFO test.py line 305 38140] <<<<<<<<<<<<<<<<< End Evaluation <<<<<<<<<<<<<<<<<

[LiyaoTang]

Hi, thanks for your interest.

Based on the result, it seems the model is randomly guessing.

There are several potentials. For example, the model checkpoint is not properly restored, or there might be some problems with the project setup.

Could you let me know how you run the test and if you train your own model?

Best.