+----------------------+------+-------+--------+-------+
| class | gts | dets | recall | ap |
+----------------------+------+-------+--------+-------+
| small_vehicle | 420 | 3460 | 0.352 | 0.110 |
| bus | 585 | 3566 | 0.803 | 0.528 |
| medium_vehicle | 1567 | 13029 | 0.705 | 0.311 |
| large_vehicle | 988 | 5344 | 0.743 | 0.500 |
| double_trailer_truck | 103 | 654 | 0.709 | 0.549 |
| small_aircraft | 485 | 1293 | 0.868 | 0.778 |
| large_aircraft | 355 | 583 | 0.972 | 0.890 |
| small_vessel | 334 | 918 | 0.455 | 0.226 |
| medium_vessel | 669 | 1964 | 0.764 | 0.608 |
| large_vessel | 366 | 750 | 0.842 | 0.751 |
| heavy_equipment | 362 | 4471 | 0.558 | 0.209 |
| container | 607 | 4350 | 0.458 | 0.231 |
| pylon | 479 | 1466 | 0.582 | 0.472 |
+----------------------+------+-------+--------+-------+
| mAP | | | | 0.474 |
+----------------------+------+-------+--------+-------+
2023-04-20 04:49:00,465 - mmdet - INFO - Epoch(val) [10][1088] mAP: 0.4741
map 0.4741 even after following training for 10 epochs and using same code as yours is the map this low expected from default code(in validation dataset)?
Posted by: prak @ April 20, 2023, 5:10 a.m.PS: Configuration changes in the baseline model notebook for results reproducibility(used every info in this thread)
Posted by: prak @ April 20, 2023, 5:11 a.m.I had the same issue for a long time. In my case it was because I did not normalize correctly i preprocessing during training
Posted by: yaelmaor @ April 20, 2023, 6:31 a.m.Hi,
Please verify whether your score range is between 0 to 1 or 0 to 100. (It does not look the same as the competition's output).
Neta,
MAFAT Challenge Team
Hi,
Thanks for replying. I would like to whether scoring range is a parameter. Coz I have just ran the code as it is. Secondly, I noticed my scores to be 0.55 is it expected to be that low using baseline?
Hi,
The baseline score should be about 33.84 on the public test set.
Neta,
MAFAT Challenge Team
"I had the same issue for a long time. In my case it was because I did not normalize correctly i preprocessing during training". Hey could you tell me whats the normalization value used for training? Was it mean=[795.0878, 795.0878, 795.0878], std=[460.0955, 460.0955, 460.0955], to_rgb=True)?
Posted by: prak @ April 21, 2023, 4:41 p.m.