Commit Graph

81 Commits

Author SHA1 Message Date
Guillermo García b1fb6fa33d train.py resume argument as store_true
Default is false.

If want to resume, call train.py --resume
2018-12-05 16:57:16 +01:00
Guillermo García c807c16b79 Fix argument parser bad practice
Keep parsing inside __main__ block and call methods with arguments

Add double -- for long argument names (- reserved for shortcuts)
2018-12-05 16:57:16 +01:00
Guillermo García 5a566454f5 Extract seed and cuda initialization utils 2018-12-05 11:55:27 +01:00
Glenn Jocher be8603b2dd updates 2018-12-04 19:17:03 +01:00
Glenn Jocher 10cca39934 updates 2018-12-03 21:08:45 +01:00
Glenn Jocher 43d74fd840 updates 2018-12-03 15:42:10 +01:00
Glenn Jocher b64620cf75 updates 2018-12-03 14:12:46 +01:00
Glenn Jocher 5843c41dfc add multi_scale support 2018-12-03 14:05:50 +01:00
Glenn Jocher b0c0182062 updates 2018-11-30 11:57:14 +01:00
Glenn Jocher 0240ac44f6 updates 2018-11-30 11:56:38 +01:00
Glenn Jocher bd649f241f updates 2018-11-29 12:12:48 +01:00
Glenn Jocher 5a0575af3a updates 2018-11-27 18:43:46 +01:00
Glenn Jocher ab9ee6aa9a updates 2018-11-23 19:45:39 +01:00
Glenn Jocher bf66656b4e updates 2018-11-23 15:34:49 +01:00
Glenn Jocher 6e825acb72 updates 2018-11-23 15:32:41 +01:00
Glenn Jocher 57f2b3f6d7 updates 2018-11-22 16:42:58 +01:00
Glenn Jocher 06579775a3
Merge pull request #37 from nirbenz/nirbenz
Fixed NMS bug causing big CPU usage.
2018-11-22 16:29:28 +01:00
Glenn Jocher b9d87be318 updates 2018-11-22 14:54:52 +01:00
Nir Ben-Zvi d41f85702d Fixed NMS bug causing big CPU usage. Note that using 'cross_class_nms' still takes a huge amount of time and should be fixed somehow. 2018-11-22 15:36:14 +02:00
Glenn Jocher f18f288990 updates 2018-11-22 14:29:50 +01:00
Glenn Jocher db515a4535 updates 2018-11-22 14:14:19 +01:00
Glenn Jocher 809667404f updates 2018-11-22 13:52:22 +01:00
Glenn Jocher 8ac8d0a382 updates 2018-11-21 19:22:35 +01:00
Glenn Jocher dae9b8f4b5 updates 2018-11-21 19:22:01 +01:00
Glenn Jocher 1415a798fe updates 2018-11-17 12:54:44 +01:00
Glenn Jocher dd7c3d2455 updates 2018-11-17 12:51:22 +01:00
Glenn Jocher 45c5567723 mAP recorded during training 2018-11-14 15:14:41 +00:00
Glenn Jocher 9dbc3ec1c4 updates 2018-11-13 11:20:01 +00:00
Glenn Jocher 34bc12d2ad updates 2018-11-11 18:58:41 +01:00
Glenn Jocher 538e5741c6 updates 2018-11-08 12:42:39 +01:00
Glenn Jocher a6d69cefe0 updates 2018-11-08 12:26:23 +01:00
Glenn Jocher 8d7660b438 updates 2018-11-07 15:17:00 +01:00
Glenn Jocher edfad8095d updates 2018-11-05 23:34:26 +01:00
Glenn Jocher 6e5da1ce27 updates 2018-11-05 23:32:36 +01:00
Glenn Jocher 3afb29ad48 update multi-scale training 2018-11-05 23:20:45 +01:00
Glenn Jocher 77469a5268 update multi-scale training 2018-11-05 23:17:53 +01:00
Glenn Jocher 0c89122aab updates 2018-11-04 18:19:07 +01:00
Glenn Jocher 332fe002b3 rename /checkpoints to /weights 2018-10-30 14:58:26 +01:00
Glenn Jocher 0ae90d0fb7 rename /checkpoints to /weights 2018-10-27 00:42:34 +02:00
Glenn Jocher 24a41972cb BCE to CE lconf + batch size 16 2018-10-11 17:43:34 +02:00
Glenn Jocher d336e0053d per-class mAP report 2018-10-10 17:07:21 +02:00
Glenn Jocher f79e7ffa76 updates 2018-10-10 16:16:17 +02:00
Glenn Jocher 0cc8f2be01 clean up train.py 2018-10-09 19:22:33 +02:00
Glenn Jocher 07ac4fef8d create step lr schedule 2018-10-05 17:01:07 +02:00
Glenn Jocher c01b8e6b7c updates 2018-10-05 16:38:59 +02:00
Glenn Jocher 8a87521044 P and R conf thresh to 0.5 2018-10-05 16:01:27 +02:00
Glenn Jocher bce94f6ade corrected numpy printoptions 2018-10-03 13:55:56 +02:00
Glenn Jocher 5d402ad31a reapply yolo width and height 2018-09-23 22:41:36 +02:00
Glenn Jocher cf9b4cfa52 update loss components 2018-09-23 22:25:23 +02:00
Glenn Jocher a722601ef6 Adam to SGD with burn-in 2018-09-20 18:03:19 +02:00