minor logging stuff

Author: antoinefalisse

main.py

@@ -13,6 +13,9 @@
 import yaml
 import traceback
 
+import logging
+logging.basicConfig(level=logging.INFO)
+
 from utils import importMetadata, loadCameraParameters, getVideoExtension
 from utils import getDataDirectory, getOpenPoseDirectory, getMMposeDirectory
 from utilsChecker import saveCameraParameters
@@ -203,16 +206,15 @@ def main(sessionName, trialName, trial_id, camerasToUse=['all'],
             # Intrinsics and extrinsics already exist for this session.
             if os.path.exists(
                     os.path.join(camDir,"cameraIntrinsicsExtrinsics.pickle")):
-                print("Load extrinsics for {} - already existing".format(
+                logging.info("Load extrinsics for {} - already existing".format(
                     camName))
                 CamParams = loadCameraParameters(
                     os.path.join(camDir, "cameraIntrinsicsExtrinsics.pickle"))
                 loadedCamParams[camName] = True
 
             # Extrinsics do not exist for this session.
             else:
-                print("Compute extrinsics for {} - not yet existing".format(
-                    camName))
+                logging.info("Compute extrinsics for {} - not yet existing".format(camName))
                 # Intrinsics ##################################################
                 # Intrinsics directories.
                 intrinsicDir = os.path.join(baseDir, 'CameraIntrinsics',
@@ -408,7 +410,7 @@ def main(sessionName, trialName, trial_id, camerasToUse=['all'],
     if runMarkerAugmentation:
         os.makedirs(postAugmentationDir, exist_ok=True)    
         augmenterDir = os.path.join(baseDir, "MarkerAugmenter")
-        print('Augmenting marker set')
+        logging.info('Augmenting marker set')
         try:
             vertical_offset = augmentTRC(
                 pathOutputFiles[trialName],sessionMetadata['mass_kg'], 
@@ -480,11 +482,11 @@ def main(sessionName, trialName, trial_id, camerasToUse=['all'],
                             thresholdTime=0.1, removeRoot=True)
                         success = True
                     except Exception as e:
-                        print(f"Attempt with thresholdPosition {thresholdPosition} failed: {e}")
+                        logging.info(f"Attempt identifying scaling time range with thresholdPosition {thresholdPosition} failed: {e}")
                         thresholdPosition += increment  # Increase the threshold for the next iteration
 
                 # Run scale tool.
-                print('Running Scaling')
+                logging.info('Running Scaling')
                 pathScaledModel = runScaleTool(
                     pathGenericSetupFile4Scaling, pathGenericModel4Scaling,
                     sessionMetadata['mass_kg'], pathTRCFile4Scaling, 
@@ -522,7 +524,7 @@ def main(sessionName, trialName, trial_id, camerasToUse=['all'],
                 # Path TRC file.
                 pathTRCFile4IK = pathAugmentedOutputFiles[trialName]
                 # Run IK tool. 
-                print('Running Inverse Kinematics')
+                logging.info('Running Inverse Kinematics')
                 try:
                     pathOutputIK = runIKTool(
                         pathGenericSetupFile4IK, pathScaledModel, 

utilsAPI.py

@@ -7,8 +7,6 @@
 import os
 import boto3
 import requests
-import logging
-logging.basicConfig(level=logging.INFO)
 
 from decouple import config
 from datetime import datetime, timedelta
@@ -51,18 +49,13 @@ def getASInstance():
         # Check if the ECS_CONTAINER_METADATA_FILE environment variable exists
         ecs_metadata_file = os.getenv('ECS_CONTAINER_METADATA_FILE')
         if ecs_metadata_file:
-            logging.info(f"ECS_CONTAINER_METADATA_FILE is set to: {ecs_metadata_file}")
             if os.path.isfile(ecs_metadata_file):
-                logging.info("Metadata file exists.")
                 return True
             else:
-                logging.warning("Metadata file does not exist at the specified path.")
                 return False
         else:
-            logging.info("ECS_CONTAINER_METADATA_FILE is not set.")
             return False
     except Exception as e:
-        logging.error(f"Error occurred while checking ECS_CONTAINER_METADATA_FILE: {e}")
         return False
 
 def get_metric_average(namespace, metric_name, start_time, end_time, period):
@@ -100,9 +93,7 @@ def get_number_of_pending_trials(period=60):
 
     if stats['Datapoints']:
         average = stats['Datapoints'][0]['Average']
-        logging.info(f"Average value of '{metric_name}' over the last minute: {average}")
     else:
-        logging.info("No data points found.")
         # Maybe raise an exception or do nothing to have control-loop retry this call later
         return None
 
@@ -131,8 +122,5 @@ def set_instance_protection(instance_id, asg_name, protect):
 
 def unprotect_current_instance():
     instance_id = get_instance_id()
-    logging.info("Instance ID: " + instance_id)
     asg_name = get_auto_scaling_group_name(instance_id)
-    logging.info("Auto Scaling Group Name: " + asg_name)
     set_instance_protection(instance_id, asg_name, protect=False)
-    logging.info("Done unprotecting instance.")

utilsAugmenter.py

@@ -45,7 +45,7 @@ def augmentTRC(pathInputTRCFile, subject_mass, subject_height,
         augmenterModelType_all = [augmenterModelType_lower, augmenterModelType_upper]
         feature_markers_all = [feature_markers_lower, feature_markers_upper]
         response_markers_all = [response_markers_lower, response_markers_upper]
-    print('Using augmenter model: {}'.format(augmenter_model))
+    # print('Using augmenter model: {}'.format(augmenter_model))
 
     # %% Process data.
     # Import TRC file
@@ -112,7 +112,7 @@ def augmentTRC(pathInputTRCFile, subject_mass, subject_height,
         json_file.close()
         model = tf.keras.models.model_from_json(pretrainedModel_json)
         model.load_weights(os.path.join(augmenterModelDir, "weights.h5"))  
-        outputs = model.predict(inputs)
+        outputs = model.predict(inputs, verbose=2)
 
         # %% Post-process outputs.
         # Step 1: Reshape if necessary (eg, LSTM)

utilsDetector.py

@@ -7,9 +7,6 @@
 import sys
 import time
 
-import logging
-logging.basicConfig(level=logging.INFO)
-
 from decouple import config
 
 from utils import getOpenPoseMarkerNames, getMMposeMarkerNames, getVideoExtension
@@ -311,21 +308,17 @@ def runMMposeVideo(
             try:
                 # wait until the video is processed (i.e. until the video is removed -- then json should be ready)
                 start = time.time()
-                logging.info("Processing mmpose. Before waiting for video to be processed.")
                 while True:
                     if not os.path.isfile(vid_path):
-                        logging.info("Processing mmpose. Video not available. Break.")
                         break
 
                     if start + 60*60 < time.time():
                         raise Exception("Pose detection timed out. This is unlikely to be your fault, please report this issue on the forum. You can proceed with your data collection (videos are uploaded to the server) and later reprocess errored trials.", 'timeout - hrnet')
 
                     time.sleep(0.1)
 
-                logging.info("Processing mmpose. Before copying.")
                 # copy /data/output to pathOutputPkl
                 os.system("cp /data/output_mmpose/* {pathOutputPkl}/".format(pathOutputPkl=pathOutputPkl))        
-                logging.info("Processing mmpose. After copying.")
                 pkl_path_tmp = os.path.join(pathOutputPkl, 'human.pkl')            
                 os.rename(pkl_path_tmp, pklPath)
 

utilsMMpose.py

@@ -2,7 +2,7 @@
 import pickle
 import torch
 
-from tqdm import tqdm
+# from tqdm import tqdm
 from mmpose_utils import process_mmdet_results, frame_iter, concat, convert_instance_to_frame
 try:
     from mmdet.apis import inference_detector, init_detector
@@ -47,7 +47,8 @@ def detection_inference(model_config, model_ckpt, video_path, bbox_path,
 
     output = []
     nFrames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
-    for img in tqdm(frame_iter(cap), total=nFrames):
+    # for img in tqdm(frame_iter(cap), total=nFrames):
+    for img in frame_iter(cap):
         # test a single image, the resulting box is (x1, y1, x2, y2)
         mmdet_results = inference_detector(det_model, img)
 
@@ -87,7 +88,8 @@ def pose_inference(model_config, model_ckpt, video_path, bbox_path, pkl_path,
     # run pose inference
     print("Running pose inference...")
     instances = []
-    for batch in tqdm(dataloader):
+    # for batch in tqdm(dataloader):
+    for batch in dataloader:
         batch['img'] = batch['img'].to(device)
         batch['img_metas'] = [img_metas[0] for img_metas in batch['img_metas'].data]
         with torch.no_grad():
@@ -122,7 +124,8 @@ def pose_inference(model_config, model_ckpt, video_path, bbox_path, pkl_path,
         dataset = model.cfg.data.test.type
         dataset_info_d = get_dataset_info()
         dataset_info = DatasetInfo(dataset_info_d[dataset])
-        for pose_results, img in tqdm(zip(results, frame_iter(cap))):
+        # for pose_results, img in tqdm(zip(results, frame_iter(cap))):
+        for pose_results, img in zip(results, frame_iter(cap)):        
             for instance in pose_results:
                 instance['keypoints'] = instance['preds_with_flip']
             vis_img = vis_pose_tracking_result(model, img, pose_results,