API tools faq
paste
Advertisement
flyxtop

train log

flyxtop
Jul 28th, 2016
213
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 17.11 KB | None | 0 0
raw download clone embed print report
  1. + set -e
  2. + export PYTHONUNBUFFERED=True
  3. + PYTHONUNBUFFERED=True
  4. + GPU_ID=0
  5. + NET=ZF
  6. + NET_lc=zf
  7. + array=($@)
  8. + len=9
  9. + EXTRA_ARGS='--set EXP_DIR foobar RNG_SEED 42 TRAIN.SCALES [400,500,600,700]'
  10. + EXTRA_ARGS_SLUG='--set_EXP_DIR_foobar_RNG_SEED_42_TRAIN.SCALES_[400,500,600,700]'
  11. ++ date +%Y-%m-%d_%H-%M-%S
  12. + LOG='experiments/logs/faster_rcnn_alt_opt_ZF_--set_EXP_DIR_foobar_RNG_SEED_42_TRAIN.SCALES_[400,500,600,700].txt.2016-07-29_00-04-50'
  13. + exec
  14. ++ tee -a 'experiments/logs/faster_rcnn_alt_opt_ZF_--set_EXP_DIR_foobar_RNG_SEED_42_TRAIN.SCALES_[400,500,600,700].txt.2016-07-29_00-04-50'
  15. + echo Logging output to 'experiments/logs/faster_rcnn_alt_opt_ZF_--set_EXP_DIR_foobar_RNG_SEED_42_TRAIN.SCALES_[400,500,600,700].txt.2016-07-29_00-04-50'
  16. Logging output to experiments/logs/faster_rcnn_alt_opt_ZF_--set_EXP_DIR_foobar_RNG_SEED_42_TRAIN.SCALES_[400,500,600,700].txt.2016-07-29_00-04-50
  17. + ./tools/train_faster_rcnn_alt_opt.py --gpu 0 --net_name ZF --weights data/imagenet_models/ZF.v2.caffemodel --imdb voc_2007_trainval --cfg experiments/cfgs/faster_rcnn_alt_opt.yml --set EXP_DIR foobar RNG_SEED 42 TRAIN.SCALES '[400,500,600,700]'
  18. Called with args:
  19. Namespace(cfg_file='experiments/cfgs/faster_rcnn_alt_opt.yml', gpu_id=0, imdb_name='voc_2007_trainval', net_name='ZF', pretrained_model='data/imagenet_models/ZF.v2.caffemodel', set_cfgs=['EXP_DIR', 'foobar', 'RNG_SEED', '42', 'TRAIN.SCALES', '[400,500,600,700]'])
  20. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  21. Stage 1 RPN, init from ImageNet model
  22. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  23. Init model: data/imagenet_models/ZF.v2.caffemodel
  24. Using config:
  25. {'DEDUP_BOXES': 0.0625,
  26. 'EPS': 1e-14,
  27. 'EXP_DIR': 'foobar',
  28. 'GPU_ID': 0,
  29. 'PIXEL_MEANS': array([[[ 102.9801, 115.9465, 122.7717]]]),
  30. 'RNG_SEED': 42,
  31. 'ROOT_DIR': '/home/ckim/Neuro/py-faster-rcnn',
  32. 'TEST': {'BBOX_REG': True,
  33. 'HAS_RPN': True,
  34. 'MAX_SIZE': 1000,
  35. 'NMS': 0.3,
  36. 'PROPOSAL_METHOD': 'selective_search',
  37. 'RPN_MIN_SIZE': 16,
  38. 'RPN_NMS_THRESH': 0.7,
  39. 'RPN_POST_NMS_TOP_N': 300,
  40. 'RPN_PRE_NMS_TOP_N': 6000,
  41. 'SCALES': [600],
  42. 'SVM': False},
  43. 'TRAIN': {'ASPECT_GROUPING': True,
  44. 'BATCH_SIZE': 128,
  45. 'BBOX_INSIDE_WEIGHTS': [1.0, 1.0, 1.0, 1.0],
  46. 'BBOX_NORMALIZE_MEANS': [0.0, 0.0, 0.0, 0.0],
  47. 'BBOX_NORMALIZE_STDS': [0.1, 0.1, 0.2, 0.2],
  48. 'BBOX_NORMALIZE_TARGETS': True,
  49. 'BBOX_NORMALIZE_TARGETS_PRECOMPUTED': False,
  50. 'BBOX_REG': False,
  51. 'BBOX_THRESH': 0.5,
  52. 'BG_THRESH_HI': 0.5,
  53. 'BG_THRESH_LO': 0.1,
  54. 'FG_FRACTION': 0.25,
  55. 'FG_THRESH': 0.5,
  56. 'HAS_RPN': True,
  57. 'IMS_PER_BATCH': 1,
  58. 'MAX_SIZE': 1000,
  59. 'PROPOSAL_METHOD': 'gt',
  60. 'RPN_BATCHSIZE': 256,
  61. 'RPN_BBOX_INSIDE_WEIGHTS': [1.0, 1.0, 1.0, 1.0],
  62. 'RPN_CLOBBER_POSITIVES': False,
  63. 'RPN_FG_FRACTION': 0.5,
  64. 'RPN_MIN_SIZE': 16,
  65. 'RPN_NEGATIVE_OVERLAP': 0.3,
  66. 'RPN_NMS_THRESH': 0.7,
  67. 'RPN_POSITIVE_OVERLAP': 0.7,
  68. 'RPN_POSITIVE_WEIGHT': -1.0,
  69. 'RPN_POST_NMS_TOP_N': 2000,
  70. 'RPN_PRE_NMS_TOP_N': 12000,
  71. 'SCALES': [400, 500, 600, 700],
  72. 'SNAPSHOT_INFIX': 'stage1',
  73. 'SNAPSHOT_ITERS': 10000,
  74. 'USE_FLIPPED': True,
  75. 'USE_PREFETCH': False},
  76. 'USE_GPU_NMS': True}
  77. Loaded dataset `voc_2007_trainval` for training
  78. Set proposal method: gt
  79. Appending horizontally-flipped training examples...
  80. voc_2007_trainval gt roidb loaded from /home/ckim/Neuro/py-faster-rcnn/data/cache/voc_2007_trainval_gt_roidb.pkl
  81. done
  82. Preparing training data...
  83. done
  84. roidb len: 10022
  85. Output will be saved to `/home/ckim/Neuro/py-faster-rcnn/output/default/voc_2007_trainval`
  86. starting Parse..
  87. WARNING: Logging before InitGoogleLogging() is written to STDERR
  88. I0729 00:04:54.295856 29609 solver.cpp:54] Initializing solver from parameters:
  89. train_net: "models/ZF/faster_rcnn_alt_opt/stage1_rpn_train.pt"
  90. base_lr: 0.001
  91. display: 20
  92. lr_policy: "step"
  93. gamma: 0.1
  94. momentum: 0.9
  95. weight_decay: 0.0005
  96. stepsize: 60000
  97. snapshot: 0
  98. snapshot_prefix: "zf_rpn"
  99. average_loss: 100
  100. I0729 00:04:54.295915 29609 solver.cpp:86] Creating training net from train_net file: models/ZF/faster_rcnn_alt_opt/stage1_rpn_train.pt
  101. [libprotobuf INFO google/protobuf/text_format.cc:1155] This one?
  102. starting Parse..
  103. I0729 00:04:54.296700 29609 net.cpp:59] Initializing net from parameters X :
  104. name: "ZF"
  105. state {
  106. phase: TRAIN
  107. }
  108. layer {
  109. name: "input-data"
  110. type: "Python"
  111. top: "data"
  112. top: "im_info"
  113. top: "gt_boxes"
  114. python_param {
  115. module: "roi_data_layer.layer"
  116. layer: "RoIDataLayer"
  117. param_str: "\'num_classes\': 21"
  118. }
  119. }
  120. layer {
  121. name: "conv1"
  122. type: "Convolution"
  123. bottom: "data"
  124. top: "conv1"
  125. param {
  126. lr_mult: 1
  127. }
  128. param {
  129. lr_mult: 2
  130. }
  131. convolution_param {
  132. num_output: 96
  133. pad: 3
  134. kernel_size: 7
  135. stride: 2
  136. }
  137. }
  138. layer {
  139. name: "relu1"
  140. type: "ReLU"
  141. bottom: "conv1"
  142. top: "conv1"
  143. }
  144. layer {
  145. name: "norm1"
  146. type: "LRN"
  147. bottom: "conv1"
  148. top: "norm1"
  149. lrn_param {
  150. local_size: 3
  151. alpha: 5e-05
  152. beta: 0.75
  153. norm_region: WITHIN_CHANNEL
  154. }
  155. }
  156. layer {
  157. name: "pool1"
  158. type: "Pooling"
  159. bottom: "norm1"
  160. top: "pool1"
  161. pooling_param {
  162. pool: MAX
  163. kernel_size: 3
  164. stride: 2
  165. pad: 1
  166. }
  167. }
  168. layer {
  169. name: "conv2"
  170. type: "Convolution"
  171. bottom: "pool1"
  172. top: "conv2"
  173. param {
  174. lr_mult: 1
  175. }
  176. param {
  177. lr_mult: 2
  178. }
  179. convolution_param {
  180. num_output: 256
  181. pad: 2
  182. kernel_size: 5
  183. stride: 2
  184. }
  185. }
  186. layer {
  187. name: "relu2"
  188. type: "ReLU"
  189. bottom: "conv2"
  190. top: "conv2"
  191. }
  192. layer {
  193. name: "norm2"
  194. type: "LRN"
  195. bottom: "conv2"
  196. top: "norm2"
  197. lrn_param {
  198. local_size: 3
  199. alpha: 5e-05
  200. beta: 0.75
  201. norm_region: WITHIN_CHANNEL
  202. }
  203. }
  204. layer {
  205. name: "pool2"
  206. type: "Pooling"
  207. bottom: "norm2"
  208. top: "pool2"
  209. pooling_param {
  210. pool: MAX
  211. kernel_size: 3
  212. stride: 2
  213. pad: 1
  214. }
  215. }
  216. layer {
  217. name: "conv3"
  218. type: "Convolution"
  219. bottom: "pool2"
  220. top: "conv3"
  221. param {
  222. lr_mult: 1
  223. }
  224. param {
  225. lr_mult: 2
  226. }
  227. convolution_param {
  228. num_output: 384
  229. pad: 1
  230. kernel_size: 3
  231. stride: 1
  232. }
  233. }
  234. layer {
  235. name: "relu3"
  236. type: "ReLU"
  237. bottom: "conv3"
  238. top: "conv3"
  239. }
  240. layer {
  241. name: "conv4"
  242. type: "Convolution"
  243. bottom: "conv3"
  244. top: "conv4"
  245. param {
  246. lr_mult: 1
  247. }
  248. param {
  249. lr_mult: 2
  250. }
  251. convolution_param {
  252. num_output: 384
  253. pad: 1
  254. kernel_size: 3
  255. stride: 1
  256. }
  257. }
  258. layer {
  259. name: "relu4"
  260. type: "ReLU"
  261. bottom: "conv4"
  262. top: "conv4"
  263. }
  264. layer {
  265. name: "conv5"
  266. type: "Convolution"
  267. bottom: "conv4"
  268. top: "conv5"
  269. param {
  270. lr_mult: 1
  271. }
  272. param {
  273. lr_mult: 2
  274. }
  275. convolution_param {
  276. num_output: 256
  277. pad: 1
  278. kernel_size: 3
  279. stride: 1
  280. }
  281. }
  282. layer {
  283. name: "relu5"
  284. type: "ReLU"
  285. bottom: "conv5"
  286. top: "conv5"
  287. }
  288. layer {
  289. name: "rpn_conv1"
  290. type: "Convolution"
  291. bottom: "conv5"
  292. top: "rpn_conv1"
  293. param {
  294. lr_mult: 1
  295. }
  296. param {
  297. lr_mult: 2
  298. }
  299. convolution_param {
  300. num_output: 256
  301. pad: 1
  302. kernel_size: 3
  303. stride: 1
  304. weight_filler {
  305. type: "gaussian"
  306. std: 0.01
  307. }
  308. bias_filler {
  309. type: "constant"
  310. value: 0
  311. }
  312. }
  313. }
  314. layer {
  315. name: "rpn_relu1"
  316. type: "ReLU"
  317. bottom: "rpn_conv1"
  318. top: "rpn_conv1"
  319. }
  320. layer {
  321. name: "rpn_cls_score"
  322. type: "Convolution"
  323. bottom: "rpn_conv1"
  324. top: "rpn_cls_score"
  325. param {
  326. lr_mult: 1
  327. }
  328. param {
  329. lr_mult: 2
  330. }
  331. convolution_param {
  332. num_output: 18
  333. pad: 0
  334. kernel_size: 1
  335. stride: 1
  336. weight_filler {
  337. type: "gaussian"
  338. std: 0.01
  339. }
  340. bias_filler {
  341. type: "constant"
  342. value: 0
  343. }
  344. }
  345. }
  346. layer {
  347. name: "rpn_bbox_pred"
  348. type: "Convolution"
  349. bottom: "rpn_conv1"
  350. top: "rpn_bbox_pred"
  351. param {
  352. lr_mult: 1
  353. }
  354. param {
  355. lr_mult: 2
  356. }
  357. convolution_param {
  358. num_output: 36
  359. pad: 0
  360. kernel_size: 1
  361. stride: 1
  362. weight_filler {
  363. type: "gaussian"
  364. std: 0.01
  365. }
  366. bias_filler {
  367. type: "constant"
  368. value: 0
  369. }
  370. }
  371. }
  372. layer {
  373. name: "rpn_cls_score_reshape"
  374. type: "Reshape"
  375. bottom: "rpn_cls_score"
  376. top: "rpn_cls_score_reshape"
  377. reshape_param {
  378. shape {
  379. dim: 0
  380. dim: 2
  381. dim: -1
  382. dim: 0
  383. }
  384. }
  385. }
  386. layer {
  387. name: "rpn-data"
  388. type: "Python"
  389. bottom: "rpn_cls_score"
  390. bottom: "gt_boxes"
  391. bottom: "im_info"
  392. bottom: "data"
  393. top: "rpn_labels"
  394. top: "rpn_bbox_targets"
  395. top: "rpn_bbox_inside_weights"
  396. top: "rpn_bbox_outside_weights"
  397. python_param {
  398. module: "rpn.anchor_target_layer"
  399. layer: "AnchorTargetLayer"
  400. param_str: "\'feat_stride\': 16"
  401. }
  402. }
  403. layer {
  404. name: "rpn_loss_cls"
  405. type: "SoftmaxWithLoss"
  406. bottom: "rpn_cls_score_reshape"
  407. bottom: "rpn_labels"
  408. top: "rpn_cls_loss"
  409. loss_weight: 1
  410. propagate_down: true
  411. propagate_down: false
  412. loss_param {
  413. ignore_label: -1
  414. normalize: true
  415. }
  416. }
  417. layer {
  418. name: "rpn_loss_bbox"
  419. type: "SmoothL1Loss"
  420. bottom: "rpn_bbox_pred"
  421. bottom: "rpn_bbox_targets"
  422. bottom: "rpn_bbox_inside_weights"
  423. bottom: "rpn_bbox_outside_weights"
  424. top: "rpn_loss_bbox"
  425. loss_weight: 1
  426. smooth_l1_loss_param {
  427. sigma: 3
  428. }
  429. }
  430. layer {
  431. name: "dummy_roi_pool_conv5"
  432. type: "DummyData"
  433. top: "dummy_roi_pool_conv5"
  434. dummy_data_param {
  435. data_filler {
  436. type: "gaussian"
  437. std: 0.01
  438. }
  439. shape {
  440. dim: 1
  441. dim: 9216
  442. }
  443. }
  444. }
  445. layer {
  446. name: "fc6"
  447. type: "InnerProduct"
  448. bottom: "dummy_roi_pool_conv5"
  449. top: "fc6"
  450. param {
  451. lr_mult: 0
  452. decay_mult: 0
  453. }
  454. param {
  455. lr_mult: 0
  456. decay_mult: 0
  457. }
  458. inner_product_param {
  459. num_output: 4096
  460. }
  461. }
  462. layer {
  463. name: "relu6"
  464. type: "ReLU"
  465. bottom: "fc6"
  466. top: "fc6"
  467. }
  468. layer {
  469. name: "fc7"
  470. type: "InnerProduct"
  471. bottom: "fc6"
  472. top: "fc7"
  473. param {
  474. lr_mult: 0
  475. decay_mult: 0
  476. }
  477. param {
  478. lr_mult: 0
  479. decay_mult: 0
  480. }
  481. inner_product_param {
  482. num_output: 4096
  483. }
  484. }
  485. layer {
  486. name: "silence_fc7"
  487. type: "Silence"
  488. bottom: "fc7"
  489. }
  490. last_loss = 1
  491. last_loss = 1
  492. state {
  493. phase: TRAIN
  494. }
  495. layer {
  496. name: "input-data"
  497. type: "Python"
  498. top: "data"
  499. top: "im_info"
  500. top: "gt_boxes"
  501. python_param {
  502. module: "roi_data_layer.layer"
  503. layer: "RoIDataLayer"
  504. param_str: "\'num_classes\': 21"
  505. }
  506. }
  507. layer {
  508. name: "data_input-data_0_split"
  509. type: "Split"
  510. bottom: "data"
  511. top: "data_input-data_0_split_0"
  512. top: "data_input-data_0_split_1"
  513. }
  514. layer {
  515. name: "conv1"
  516. type: "Convolution"
  517. bottom: "data_input-data_0_split_0"
  518. top: "conv1"
  519. param {
  520. lr_mult: 1
  521. }
  522. param {
  523. lr_mult: 2
  524. }
  525. convolution_param {
  526. num_output: 96
  527. pad: 3
  528. kernel_size: 7
  529. stride: 2
  530. }
  531. }
  532. layer {
  533. name: "relu1"
  534. type: "ReLU"
  535. bottom: "conv1"
  536. top: "conv1"
  537. }
  538. layer {
  539. name: "norm1"
  540. type: "LRN"
  541. bottom: "conv1"
  542. top: "norm1"
  543. lrn_param {
  544. local_size: 3
  545. alpha: 5e-05
  546. beta: 0.75
  547. norm_region: WITHIN_CHANNEL
  548. }
  549. }
  550. layer {
  551. name: "pool1"
  552. type: "Pooling"
  553. bottom: "norm1"
  554. top: "pool1"
  555. pooling_param {
  556. pool: MAX
  557. kernel_size: 3
  558. stride: 2
  559. pad: 1
  560. }
  561. }
  562. layer {
  563. name: "conv2"
  564. type: "Convolution"
  565. bottom: "pool1"
  566. top: "conv2"
  567. param {
  568. lr_mult: 1
  569. }
  570. param {
  571. lr_mult: 2
  572. }
  573. convolution_param {
  574. num_output: 256
  575. pad: 2
  576. kernel_size: 5
  577. stride: 2
  578. }
  579. }
  580. layer {
  581. name: "relu2"
  582. type: "ReLU"
  583. bottom: "conv2"
  584. top: "conv2"
  585. }
  586. layer {
  587. name: "norm2"
  588. type: "LRN"
  589. bottom: "conv2"
  590. top: "norm2"
  591. lrn_param {
  592. local_size: 3
  593. alpha: 5e-05
  594. beta: 0.75
  595. norm_region: WITHIN_CHANNEL
  596. }
  597. }
  598. layer {
  599. name: "pool2"
  600. type: "Pooling"
  601. bottom: "norm2"
  602. top: "pool2"
  603. pooling_param {
  604. pool: MAX
  605. kernel_size: 3
  606. stride: 2
  607. pad: 1
  608. }
  609. }
  610. layer {
  611. name: "conv3"
  612. type: "Convolution"
  613. bottom: "pool2"
  614. top: "conv3"
  615. param {
  616. lr_mult: 1
  617. }
  618. param {
  619. lr_mult: 2
  620. }
  621. convolution_param {
  622. num_output: 384
  623. pad: 1
  624. kernel_size: 3
  625. stride: 1
  626. }
  627. }
  628. layer {
  629. name: "relu3"
  630. type: "ReLU"
  631. bottom: "conv3"
  632. top: "conv3"
  633. }
  634. layer {
  635. name: "conv4"
  636. type: "Convolution"
  637. bottom: "conv3"
  638. top: "conv4"
  639. param {
  640. lr_mult: 1
  641. }
  642. param {
  643. lr_mult: 2
  644. }
  645. convolution_param {
  646. num_output: 384
  647. pad: 1
  648. kernel_size: 3
  649. stride: 1
  650. }
  651. }
  652. layer {
  653. name: "relu4"
  654. type: "ReLU"
  655. bottom: "conv4"
  656. top: "conv4"
  657. }
  658. layer {
  659. name: "conv5"
  660. type: "Convolution"
  661. bottom: "conv4"
  662. top: "conv5"
  663. param {
  664. lr_mult: 1
  665. }
  666. param {
  667. lr_mult: 2
  668. }
  669. convolution_param {
  670. num_output: 256
  671. pad: 1
  672. kernel_size: 3
  673. stride: 1
  674. }
  675. }
  676. layer {
  677. name: "relu5"
  678. type: "ReLU"
  679. bottom: "conv5"
  680. top: "conv5"
  681. }
  682. layer {
  683. name: "rpn_conv1"
  684. type: "Convolution"
  685. bottom: "conv5"
  686. top: "rpn_conv1"
  687. param {
  688. lr_mult: 1
  689. }
  690. param {
  691. lr_mult: 2
  692. }
  693. convolution_param {
  694. num_output: 256
  695. pad: 1
  696. kernel_size: 3
  697. stride: 1
  698. weight_filler {
  699. type: "gaussian"
  700. std: 0.01
  701. }
  702. bias_filler {
  703. type: "constant"
  704. value: 0
  705. }
  706. }
  707. }
  708. layer {
  709. name: "rpn_relu1"
  710. type: "ReLU"
  711. bottom: "rpn_conv1"
  712. top: "rpn_conv1"
  713. }
  714. layer {
  715. name: "rpn_conv1_rpn_relu1_0_split"
  716. type: "Split"
  717. bottom: "rpn_conv1"
  718. top: "rpn_conv1_rpn_relu1_0_split_0"
  719. top: "rpn_conv1_rpn_relu1_0_split_1"
  720. }
  721. layer {
  722. name: "rpn_cls_score"
  723. type: "Convolution"
  724. bottom: "rpn_conv1_rpn_relu1_0_split_0"
  725. top: "rpn_cls_score"
  726. param {
  727. lr_mult: 1
  728. }
  729. param {
  730. lr_mult: 2
  731. }
  732. convolution_param {
  733. num_output: 18
  734. pad: 0
  735. kernel_size: 1
  736. stride: 1
  737. weight_filler {
  738. type: "gaussian"
  739. std: 0.01
  740. }
  741. bias_filler {
  742. type: "constant"
  743. value: 0
  744. }
  745. }
  746. }
  747. layer {
  748. name: "rpn_cls_score_rpn_cls_score_0_split"
  749. type: "Split"
  750. bottom: "rpn_cls_score"
  751. top: "rpn_cls_score_rpn_cls_score_0_split_0"
  752. top: "rpn_cls_score_rpn_cls_score_0_split_1"
  753. }
  754. layer {
  755. name: "rpn_bbox_pred"
  756. type: "Convolution"
  757. bottom: "rpn_conv1_rpn_relu1_0_split_1"
  758. top: "rpn_bbox_pred"
  759. param {
  760. lr_mult: 1
  761. }
  762. param {
  763. lr_mult: 2
  764. }
  765. convolution_param {
  766. num_output: 36
  767. pad: 0
  768. kernel_size: 1
  769. stride: 1
  770. weight_filler {
  771. type: "gaussian"
  772. std: 0.01
  773. }
  774. bias_filler {
  775. type: "constant"
  776. value: 0
  777. }
  778. }
  779. }
  780. layer {
  781. name: "rpn_cls_score_reshape"
  782. type: "Reshape"
  783. bottom: "rpn_cls_score_rpn_cls_score_0_split_0"
  784. top: "rpn_cls_score_reshape"
  785. reshape_param {
  786. shape {
  787. dim: 0
  788. dim: 2
  789. dim: -1
  790. dim: 0
  791. }
  792. }
  793. }
  794. layer {
  795. name: "rpn-data"
  796. type: "Python"
  797. bottom: "rpn_cls_score_rpn_cls_score_0_split_1"
  798. bottom: "gt_boxes"
  799. bottom: "im_info"
  800. bottom: "data_input-data_0_split_1"
  801. top: "rpn_labels"
  802. top: "rpn_bbox_targets"
  803. top: "rpn_bbox_inside_weights"
  804. top: "rpn_bbox_outside_weights"
  805. python_param {
  806. module: "rpn.anchor_target_layer"
  807. layer: "AnchorTargetLayer"
  808. param_str: "\'feat_stride\': 16"
  809. }
  810. }
  811. layer {
  812. name: "rpn_loss_cls"
  813. type: "SoftmaxWithLoss"
  814. bottom: "rpn_cls_score_reshape"
  815. bottom: "rpn_labels"
  816. top: "rpn_cls_loss"
  817. loss_weight: 1
  818. propagate_down: true
  819. propagate_down: false
  820. loss_param {
  821. ignore_label: -1
  822. normalize: true
  823. }
  824. }
  825. layer {
  826. name: "rpn_loss_bbox"
  827. type: "SmoothL1Loss"
  828. bottom: "rpn_bbox_pred"
  829. bottom: "rpn_bbox_targets"
  830. bottom: "rpn_bbox_inside_weights"
  831. bottom: "rpn_bbox_outside_weights"
  832. top: "rpn_loss_bbox"
  833. loss_weight: 1
  834. smooth_l1_loss_param {
  835. sigma: 3
  836. }
  837. }
  838. layer {
  839. name: "dummy_roi_pool_conv5"
  840. type: "DummyData"
  841. top: "dummy_roi_pool_conv5"
  842. dummy_data_param {
  843. data_filler {
  844. type: "gaussian"
  845. std: 0.01
  846. }
  847. shape {
  848. dim: 1
  849. dim: 9216
  850. }
  851. }
  852. }
  853. layer {
  854. name: "fc6"
  855. type: "InnerProduct"
  856. bottom: "dummy_roi_pool_conv5"
  857. top: "fc6"
  858. param {
  859. lr_mult: 0
  860. decay_mult: 0
  861. }
  862. param {
  863. lr_mult: 0
  864. decay_mult: 0
  865. }
  866. inner_product_param {
  867. num_output: 4096
  868. }
  869. }
  870. layer {
  871. name: "relu6"
  872. type: "ReLU"
  873. bottom: "fc6"
  874. top: "fc6"
  875. }
  876. layer {
  877. name: "fc7"
  878. type: "InnerProduct"
  879. bottom: "fc6"
  880. top: "fc7"
  881. param {
  882. lr_mult: 0
  883. decay_mult: 0
  884. }
  885. param {
  886. lr_mult: 0
  887. decay_mult: 0
  888. }
  889. inner_product_param {
  890. num_output: 4096
  891. }
  892. }
  893. layer {
  894. name: "silence_fc7"
  895. type: "Silence"
  896. bottom: "fc7"
  897. }
  898. I0729 00:04:54.297251 29609 layer_factory.hpp:76] Creating layer input-data
  899. I0729 00:04:54.297762 29609 net.cpp:140] Creating Layer input-data
  900. I0729 00:04:54.297783 29609 net.cpp:476] input-data -> data
  901. I0729 00:04:54.297806 29609 net.cpp:476] input-data -> im_info
  902. I0729 00:04:54.297821 29609 net.cpp:476] input-data -> gt_boxes
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!