python /home/admin/mtr/script_for_cron.py -j python_test3 -m 12 -a ' --short_python3 -v ' -s python_test3 -M 0 -S 0 -U 100,100,120 import MySQLdb succeeded Import error (python version) python version = 3 warning , we can't find thcl infos in json_data warning , we can't find pdt infos in json_data list_job_run_as_list : ['mask_detection', 'datou', 'CacheModelData_queries', 'CachePhotoData_queries', 'test_fork', 'prepare_maskdata', 'portfolio_queries', 'sla_mensuel'] python version used : 3 liste_fichiers : [('tests/mask_test', True, 'Test mask-detection ', 'mask_detection'), ('tests/datou_test', True, 'Datou All Test', 'datou', 'all'), ('mtr/database_queries/CacheModelData_queries', True, 'Test Cache Model Data', 'CacheModelData_queries'), ('tests/cache_photo_data_test', True, 'Test local_cache_photo ', 'CachePhotoData_queries'), ('mtr/mask_rcnn/prepare_maskdata', True, 'test prepare mask data', 'prepare_maskdata', 'all'), ('mtr/database_queries/portfolio_queries', True, 'test portfolio queries', 'portfolio_queries'), ('prod/memo/memo', True, 'SLA Mensuel', 'sla_mensuel', 'all')] #&_# BEGIN OF TEST : tests/mask_test #&_# /home/admin/workarea/git/Velours/python/tests/mask_test.py Test mask-detection python version used : 3 ############################### TEST memory used ################################ free memory at begining : begin to check gpu status inside check gpu memory l 3610 free memory gpu now : 27 run mask_detect Inside batchDatouExec : verbose : False # VR 17-11-17 : to create in DB ! Here we check the datou graph and we reorder steps ! Tree builded and cycle checked, now we need to re-order the steps ! We have currenlty an error because there is no dependence between the last step for the case tile - detect - glue We can either keep the depence of, it is better to keep an order compatible with the id of steps if we do not have sons, so a lexical order : (number_son, step_id) DONE and to test : checkNoCycle ! We are managing only one step so we do not consider checkConsistencyNbInputNbOutput ! We are managing only one step so we do not consider checkConsistencyTypeOutputInput ! List Step Type Loaded in datou : mask_detect list_input_json : [] origin BFwe have missing 0 photos in the step downloads : photo missing : [] try to delete the photos missing in DB length of list_filenames : 1 ; length of list_pids : 1 ; length of list_args : 1 time to download the photos : 0.10745930671691895 About to test input to load we should then remove the video here, and this would fix the bug of datou_current ! Calling datou_exec Inside datou_exec : verbose : False number of steps : 1 step1:mask_detect Sat May 10 00:35:28 2025 VR 17-11-17 : now, only for linear exec dependencies tree, some output goes to fill the input of the next VR 22-3-18 : now we test the dependencies tree, but keep two separate code for datou_prepare_output_input until the code is correctly tested, clean and works in both case VR 22-3-18 : but we use the first code for the first step id = -1, build in the code of datou_exec VR 22-3-18 : we should manage here the case when we are at the first step instead of building this step before datou_exec Beginning of datou step mask_detect ! save_polygon : True begin detect begin to check gpu status inside check gpu memory havn't enough memory gpu , need / 3000 l 3632 free memory gpu now : 27 wait 20 seconds l 3637 free memory gpu now : 27 max_wait_temp : 1 max_wait : 0 gpu_flag : 0 /home/admin/workarea/git/Velours/python/tests/python_tests.py:11: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses import imp 2025-05-10 00:35:51.402237: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2025-05-10 00:35:51.427274: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 3493065000 Hz 2025-05-10 00:35:51.429564: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f2408000b60 initialized for platform Host (this does not guarantee that XLA will be used). Devices: 2025-05-10 00:35:51.429631: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version 2025-05-10 00:35:51.434107: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1 2025-05-10 00:35:51.471200: W tensorflow/compiler/xla/service/platform_util.cc:210] unable to create StreamExecutor for CUDA:0: failed initializing StreamExecutor for CUDA device ordinal 0: Internal: failed call to cuDevicePrimaryCtxRetain: CUDA_ERROR_OUT_OF_MEMORY: out of memory; total memory reported: 11553341440 2025-05-10 00:35:51.471343: I tensorflow/compiler/jit/xla_gpu_device.cc:161] Ignoring visible XLA_GPU_JIT device. Device number is 0, reason: Internal: no supported devices found for platform CUDA 2025-05-10 00:35:51.472003: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: pciBusID: 0000:41:00.0 name: NVIDIA GeForce RTX 2080 Ti computeCapability: 7.5 coreClock: 1.545GHz coreCount: 68 deviceMemorySize: 10.76GiB deviceMemoryBandwidth: 573.69GiB/s 2025-05-10 00:35:51.472332: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 2025-05-10 00:35:51.479523: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10 2025-05-10 00:35:51.483026: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10 2025-05-10 00:35:51.483539: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10 2025-05-10 00:35:51.486674: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10 2025-05-10 00:35:51.487956: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10 2025-05-10 00:35:51.493647: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7 2025-05-10 00:35:51.494543: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0 2025-05-10 00:35:51.494621: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 Inside mask_sub_process Inside mask_detect Exception in mask_detect : CUDA runtime implicit initialization on GPU:0 failed. Status: out of memory we want to redo the detection max_time_sub_proc : 3600 erreur pendant la detection Useless call to update_current_state in case -12 caffe_path_current : About to save ! 1 Inside saveOutput : final : True verbose : False eke 12-6-18 : saveMask need to be cleaned for new output ! ERROR : mask output needs to be a dictionnary now ! No output to save, continue without doing anything ! save missing photos in datou_result : After save, about to update current ! datou_cur_ids : [] len(datou.list_steps) : 1 output : -12 free memory after detection : begin to check gpu status inside check gpu memory l 3610 free memory gpu now : 27 ############################### TEST detect object ################################ run mask_detect Inside batchDatouExec : verbose : False # VR 17-11-17 : to create in DB ! Here we check the datou graph and we reorder steps ! Tree builded and cycle checked, now we need to re-order the steps ! We have currenlty an error because there is no dependence between the last step for the case tile - detect - glue We can either keep the depence of, it is better to keep an order compatible with the id of steps if we do not have sons, so a lexical order : (number_son, step_id) DONE and to test : checkNoCycle ! We are managing only one step so we do not consider checkConsistencyNbInputNbOutput ! We are managing only one step so we do not consider checkConsistencyTypeOutputInput ! List Step Type Loaded in datou : mask_detect list_input_json : [] origin BFwe have missing 0 photos in the step downloads : photo missing : [] try to delete the photos missing in DB length of list_filenames : 1 ; length of list_pids : 1 ; length of list_args : 1 time to download the photos : 0.13757681846618652 About to test input to load we should then remove the video here, and this would fix the bug of datou_current ! Calling datou_exec Inside datou_exec : verbose : False number of steps : 1 step1:mask_detect Sat May 10 00:35:52 2025 VR 17-11-17 : now, only for linear exec dependencies tree, some output goes to fill the input of the next VR 22-3-18 : now we test the dependencies tree, but keep two separate code for datou_prepare_output_input until the code is correctly tested, clean and works in both case VR 22-3-18 : but we use the first code for the first step id = -1, build in the code of datou_exec VR 22-3-18 : we should manage here the case when we are at the first step instead of building this step before datou_exec Beginning of datou step mask_detect ! save_polygon : True begin detect begin to check gpu status inside check gpu memory havn't enough memory gpu , need / 3000 l 3632 free memory gpu now : 27 wait 20 seconds l 3637 free memory gpu now : 27 max_wait_temp : 1 max_wait : 0 gpu_flag : 0 2025-05-10 00:36:15.414286: I tensorflow/core/platform/cpu_feature_guard.cc:143] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2025-05-10 00:36:15.439135: I tensorflow/core/platform/profile_utils/cpu_utils.cc:102] CPU Frequency: 3493065000 Hz 2025-05-10 00:36:15.441030: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f240c000b60 initialized for platform Host (this does not guarantee that XLA will be used). Devices: 2025-05-10 00:36:15.441089: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version 2025-05-10 00:36:15.469902: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1 2025-05-10 00:36:15.510466: W tensorflow/compiler/xla/service/platform_util.cc:210] unable to create StreamExecutor for CUDA:0: failed initializing StreamExecutor for CUDA device ordinal 0: Internal: failed call to cuDevicePrimaryCtxRetain: CUDA_ERROR_OUT_OF_MEMORY: out of memory; total memory reported: 11553341440 2025-05-10 00:36:15.510645: I tensorflow/compiler/jit/xla_gpu_device.cc:161] Ignoring visible XLA_GPU_JIT device. Device number is 0, reason: Internal: no supported devices found for platform CUDA 2025-05-10 00:36:15.511459: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1561] Found device 0 with properties: pciBusID: 0000:41:00.0 name: NVIDIA GeForce RTX 2080 Ti computeCapability: 7.5 coreClock: 1.545GHz coreCount: 68 deviceMemorySize: 10.76GiB deviceMemoryBandwidth: 573.69GiB/s 2025-05-10 00:36:15.511842: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 2025-05-10 00:36:15.514867: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10 2025-05-10 00:36:15.517888: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10 2025-05-10 00:36:15.518403: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10 2025-05-10 00:36:15.521804: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10 2025-05-10 00:36:15.523115: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10 2025-05-10 00:36:15.528865: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7 2025-05-10 00:36:15.529788: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1703] Adding visible gpu devices: 0 2025-05-10 00:36:15.529867: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 max_time_sub_proc : 3600 erreur pendant la detection Useless call to update_current_state in case -12 caffe_path_current : Inside mask_sub_process Inside mask_detect Exception in mask_detect : CUDA runtime implicit initialization on GPU:0 failed. Status: out of memory we want to redo the detection About to save ! 1 Inside saveOutput : final : True verbose : False eke 12-6-18 : saveMask need to be cleaned for new output ! ERROR : mask output needs to be a dictionnary now ! No output to save, continue without doing anything ! save missing photos in datou_result : After save, about to update current ! datou_cur_ids : [] len(datou.list_steps) : 1 output : -12 ERROR : 'int' object is not subscriptable reconnect to base ! warning , we can't find thcl infos in json_data warning , we can't find pdt infos in json_data #&_# TEST FAILED #&_# : tests/mask_test #&_# Error : invalid literal for int() with base 10: "'int' object is not subscriptable" /home/admin/workarea/git/Velours/python/tests/python_tests.py refs/heads/master_6b796098f0a7c88b7d6a90fb4c0df56eec821fbf SQL :INSERT INTO MTRAdmin.monitor_sys (name, type, server, version_code, result_str, result_bool, lien , test_group ,test_name) VALUES ('python_test3','1','marlene','refs/heads/master_6b796098f0a7c88b7d6a90fb4c0df56eec821fbf','{"mask_detection": "fail"}','0','http://marlene.fotonower-preprod.com/job/2025/May/10052025/python_test3//data_2/data_log/job/2025/May/10052025/python_test3/log-python3----short_python3--v--marlene-00:35:02.txt','mask_detection','unknown'); #&_# END OF TEST #&_# : tests/mask_test #&_# #&_# BEGIN OF TEST : tests/datou_test #&_# /home/admin/workarea/git/Velours/python/tests/datou_test.py Datou All Test python version used : 3 ############################### TEST sam ################################ TEST SAM Inside batchDatouExec : verbose : True ##### chargement datou SELECT name, created_at,limit_max FROM MTRDatou.mtr_datou WHERE id=4573 SELECT mtd.id, mtdt.`type`, mtd.`param`, mtd.param_json, mtdt.nb_input, mtdt.nb_output, mtdt.prod, mtdt.is_local, mtdt.is_datou_depend, mtdt.is_photo_id_local FROM MTRDatou.mtr_datou_step mtd, MTRDatou.mtr_datou_step_types mtdt WHERE mtdt.`id`=mtd.`type` AND mtd.mtd_id=4573 SELECT mtd.id, mtd.mtd_id, mdsdt.id, mdsdt.name, mdsdt.description, msid.output_or_input, msid.data_order_id, mdsdt.type FROM MTRDatou.mtr_datou_step mtd, MTRDatou.mtr_datou_steptype_io_datatypes msid, MTRDatou.mtr_datou_step_data_types mdsdt WHERE mtd.`type`=msid.`mtr_datou_step_type` AND mtd.mtd_id= 4573 AND msid.data_type=mdsdt.id SELECT mts_id_output, id_output, mts_id_input, id_input FROM MTRDatou.mtr_datou_step_by_step WHERE mtd_id=4573 # VR 17-11-17 : to create in DB ! Here we check the datou graph and we reorder steps ! Tree builded and cycle checked, now we need to re-order the steps ! We have currenlty an error because there is no dependence between the last step for the case tile - detect - glue We can either keep the depence of, it is better to keep an order compatible with the id of steps if we do not have sons, so a lexical order : (number_son, step_id) DONE and to test : checkNoCycle ! We are managing only one step so we do not consider checkConsistencyNbInputNbOutput ! We are managing only one step so we do not consider checkConsistencyTypeOutputInput ! no param json to modify List Step Type Loaded in datou : sam list_input_json : [] ##### fin chargement datou ##### chargement data ##### Call load_data_input : nb_thread : 5 origin SELECT photo_id, url FROM MTRBack.photos ph WHERE photo_id IN (1189321094) Found this number of photos: 1 ##### Call download_photos : nb_thread : 5 begin to download photo : 1189321094 download finish for photo 1189321094 we have missing 0 photos in the step downloads : photo missing : [] try to delete the photos missing in DB ##### After download_photos length of list_filenames : 1 ; length of list_pids : 1 ; length of list_args : 1 ##### After load_data_input time to download the photos : 0.16719317436218262 #### fin chargement data Blocking on flush ? No conitnuing About to test input to load we should then remove the video here, and this would fix the bug of datou_current ! WARNING : we have an input that is not a photo, we should get rid of it Calling datou_exec Inside datou_exec : verbose : True number of steps : 1 step1:sam Sat May 10 00:36:16 2025 VR 17-11-17 : now, only for linear exec dependencies tree, some output goes to fill the input of the next VR 22-3-18 : now we test the dependencies tree, but keep two separate code for datou_prepare_output_input until the code is correctly tested, clean and works in both case VR 22-3-18 : but we use the first code for the first step id = -1, build in the code of datou_exec VR 22-3-18 : we should manage here the case when we are at the first step instead of building this step before datou_exec After prepare type args : Here we display some param of map_info ! map_filenames : {'temp/1746830175_2262503_1189321094_9626af7f95d010f2a4fd524688d4ea22_76896585.png': 1189321094} map_photo_id_path_extension : {1189321094: {'path': 'temp/1746830175_2262503_1189321094_9626af7f95d010f2a4fd524688d4ea22_76896585.png', 'extension': 'png'}} map_subphoto_mainphoto : {} Beginning of datou step sam ! pht : 4677 Inside sam : nb paths : 1 ERROR in datou_step_exec, will save and exit ! CUDA error: out of memory CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. File "/home/admin/workarea/git/Velours/python/mtr/datou/datou_lib.py", line 2329, in datou_exec output = datou_step_exec(sNext, args, cache, context, map_info, verbose, mtr_user_id) File "/home/admin/workarea/git/Velours/python/mtr/datou/datou_lib.py", line 2430, in datou_step_exec return lib_process.datou_step_sam(param, json_param, args, cache, context, map_info, verbose) File "/home/admin/workarea/git/Velours/python/mtr/datou/lib_step_exec/lib_step_process.py", line 367, in datou_step_sam sam.to(device=device) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 927, in to return self._apply(convert) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 579, in _apply module._apply(fn) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 579, in _apply module._apply(fn) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 579, in _apply module._apply(fn) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 602, in _apply param_applied = fn(param) File "/home/admin/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 925, in convert return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) [1189321094] map_info['map_portfolio_photo'] : {} final : True mtd_id 4573 list_pids : [1189321094] begin to insert list_values into mtr_datou_result : length of list_values in save_final : 1 insert ignore into MTRPhoto.mtr_datou_result (mtd_id, mtr_portfolio_id,mtr_photo_id,result,result_long,result_double,hashtag_id,proba, mtr_current_id) values (%s,%s,%s,%s,%s,%s,%s,%s,%s) on duplicate key update mtr_portfolio_id = mtr_portfolio_id list_values : [('4573', None, '1189321094', "[>, , , , , 'CUDA error: out of memory\\nCUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.\\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1.']", '-1', '-1.0', '501120777', '1.0', None)] time used for this insertion : 0.021515607833862305 save_final ERROR in last step sam, CUDA error: out of memory CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. time spend for datou_step_exec : 2.855409622192383 time spend to save output : 0.023076772689819336 total time spend for step 0 : 2.878486394882202 need to delete datou_research and reload, so keep current state 1 need to delete datou_research and reload, so keep current state 1 caffe_path_current : About to save ! 2 After save, about to update current ! datou_cur_ids : [] len(datou.list_steps) : 1 output : None ERROR nb objects espect : 98 nb_objects detect : 0 ERROR sam FAILED ############################### TEST frcnn ################################ test frcnn Inside batchDatouExec : verbose : True ##### chargement datou SELECT name, created_at,limit_max FROM MTRDatou.mtr_datou WHERE id=4184 SELECT mtd.id, mtdt.`type`, mtd.`param`, mtd.param_json, mtdt.nb_input, mtdt.nb_output, mtdt.prod, mtdt.is_local, mtdt.is_datou_depend, mtdt.is_photo_id_local FROM MTRDatou.mtr_datou_step mtd, MTRDatou.mtr_datou_step_types mtdt WHERE mtdt.`id`=mtd.`type` AND mtd.mtd_id=4184 SELECT mtd.id, mtd.mtd_id, mdsdt.id, mdsdt.name, mdsdt.description, msid.output_or_input, msid.data_order_id, mdsdt.type FROM MTRDatou.mtr_datou_step mtd, MTRDatou.mtr_datou_steptype_io_datatypes msid, MTRDatou.mtr_datou_step_data_types mdsdt WHERE mtd.`type`=msid.`mtr_datou_step_type` AND mtd.mtd_id= 4184 AND msid.data_type=mdsdt.id SELECT mts_id_output, id_output, mts_id_input, id_input FROM MTRDatou.mtr_datou_step_by_step WHERE mtd_id=4184 # VR 17-11-17 : to create in DB ! Here we check the datou graph and we reorder steps ! Tree builded and cycle checked, now we need to re-order the steps ! We have currenlty an error because there is no dependence between the last step for the case tile - detect - glue We can either keep the depence of, it is better to keep an order compatible with the id of steps if we do not have sons, so a lexical order : (number_son, step_id) DONE and to test : checkNoCycle ! We are managing only one step so we do not consider checkConsistencyNbInputNbOutput ! We are managing only one step so we do not consider checkConsistencyTypeOutputInput ! no param json to modify List Step Type Loaded in datou : frcnn list_input_json : [] ##### fin chargement datou ##### chargement data ##### Call load_data_input : nb_thread : 5 origin SELECT photo_id, url FROM MTRBack.photos ph WHERE photo_id IN (917754606) Found this number of photos: 1 ##### Call download_photos : nb_thread : 5 begin to download photo : 917754606 download finish for photo 917754606 we have missing 0 photos in the step downloads : photo missing : [] try to delete the photos missing in DB ##### After download_photos length of list_filenames : 1 ; length of list_pids : 1 ; length of list_args : 1 ##### After load_data_input time to download the photos : 0.17563843727111816 #### fin chargement data Blocking on flush ? No conitnuing About to test input to load we should then remove the video here, and this would fix the bug of datou_current ! Calling datou_exec Inside datou_exec : verbose : True number of steps : 1 step1:frcnn Sat May 10 00:36:19 2025 VR 17-11-17 : now, only for linear exec dependencies tree, some output goes to fill the input of the next VR 22-3-18 : now we test the dependencies tree, but keep two separate code for datou_prepare_output_input until the code is correctly tested, clean and works in both case VR 22-3-18 : but we use the first code for the first step id = -1, build in the code of datou_exec VR 22-3-18 : we should manage here the case when we are at the first step instead of building this step before datou_exec After prepare type args : Here we display some param of map_info ! map_filenames : {'temp/1746830178_2262503_917754606_35f3c9ae49686a6be16030c6ec25c9ee.jpg': 917754606} map_photo_id_path_extension : {917754606: {'path': 'temp/1746830178_2262503_917754606_35f3c9ae49686a6be16030c6ec25c9ee.jpg', 'extension': 'jpg'}} map_subphoto_mainphoto : {} Beginning of datou step Faster rcnn ! classes : ['background', 'plaque'] pht : 4370 caffemodel_name (should be vgg16_immat_307 but not used because net loaded outside in the fonction) : {'id': 3375, 'mtr_user_id': 31, 'name': 'detection_plaque_valcor_010622', 'pb_hashtag_id': 0, 'live': b'\x00', 'list_hashtags': 'background,plaque', 'svm_portfolios_learning': '0,0', 'photo_hashtag_type': 4370, 'photo_desc_type': 5676, 'type_classification': 'caffe_faster_rcnn', 'hashtag_id_list': '0,0'} To loadFromThcl() model_param file didn't exist model_name : detection_plaque_valcor_010622 model_type : caffe_faster_rcnn list file need : ['caffemodel', 'test.prototxt'] file exist in s3 : ['caffemodel', 'test.prototxt'] file manque in s3 : [] WARNING: Logging before InitGoogleLogging() is written to STDERR E0510 00:36:20.730556 2262503 common.cpp:114] Cannot create Cublas handle. Cublas won't be available. E0510 00:36:20.738272 2262503 common.cpp:121] Cannot create Curand generator. Curand won't be available. F0510 00:36:20.749389 2262503 syncedmem.hpp:22] Check failed: error == cudaSuccess (2 vs. 0) out of memory *** Check failure stack trace: *** Command terminated by signal 6 5.57user 6.80system 0:54.68elapsed 22%CPU (0avgtext+0avgdata 1054648maxresident)k 1291608inputs+3264outputs (3027major+372271minor)pagefaults 0swaps