megadetector 10.0.7__tar.gz → 10.0.9__tar.gz

{megadetector-10.0.7/megadetector.egg-info → megadetector-10.0.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: megadetector
-Version: 10.0.7
+Version: 10.0.9
 Summary: MegaDetector is an AI model that helps conservation folks spend less time doing boring things with camera trap images.
 Author-email: Your friendly neighborhood MegaDetector team <cameratraps@lila.science>
 Maintainer-email: Your friendly neighborhood MegaDetector team <cameratraps@lila.science>

{megadetector-10.0.7 → megadetector-10.0.9}/README.md

@@ -112,7 +112,6 @@ Here are a few of the organizations that have used MegaDetector... we're only li
 * [TROPECOLNET project](https://www.anabenitezlopez.com/research/global-change-biology/tropecolnet/), Museo Nacional de Ciencias Naturales
 * [Wildlife Coexistence Lab](https://wildlife.forestry.ubc.ca/), University of British Columbia
 * [Wildlife Research](https://www.dfw.state.or.us/wildlife/research/index.asp), Oregon Department of Fish and Wildlife
-* [Wildlife Division](https://www.michigan.gov/dnr/about/contact/wildlife), Michigan Department of Natural Resources
 * [Kohl Wildlife Lab](https://kohlwildlifelab.com/), University of Georgia
 * [SPEC Lab](https://thespeclab.weebly.com/) and [Cherry Lab](https://www.ckwri.tamuk.edu/about/michael-j-cherry-phd), Caesar Kleberg Wildlife Research Institute, Texas A&M Kingsville
 * Ecology and Conservation of Amazonian Vertebrates Research Group, Federal University of Amapá
@@ -148,10 +147,11 @@ Here are a few of the organizations that have used MegaDetector... we're only li
 * [The Nature Conservancy in California](https://www.nature.org/en-us/about-us/where-we-work/united-states/california/) ([Animl platform](https://github.com/tnc-ca-geo/animl-frontend)) ([story](https://www.vision-systems.com/non-factory/environment-agriculture/article/14304433/the-nature-conservancy-brings-cameras-ai-to-invasive-species-prevention))
 * [San Diego Zoo Wildlife Alliance](https://science.sandiegozoo.org/)  ([Animl R package](https://github.com/conservationtechlab/animl))
 * [TerrOïko](https://www.terroiko.fr/) ([OCAPI platform](https://www.terroiko.fr/ocapi))
+* [Wildlife Division](https://www.michigan.gov/dnr/about/contact/wildlife), Michigan Department of Natural Resources ([blog post](https://www.michigan.gov/dnr/about/newsroom/releases/2025/08/18/dnr-researchers-to-test-trail-cameras-in-elk-survey))
 
 Also see:
 
-* The [list of MD-related GUIs, platforms, and GitHub repos](https://github.com/agentmorris/MegaDetector/blob/main/megadetector.md#is-there-a-gui) within the MegaDetector User Guide... although you can never have too many lists, so here they are in a concise comma-separated list: [Wildlife Insights](https://wildlifeinsights.org/), [Animal Detect](https://www.animaldetect.com), [TrapTagger](https://wildeyeconservation.org/trap-tagger-about/), [WildTrax](https://www.wildtrax.ca/), [Agouti](https://agouti.eu/), [Trapper](https://trapper-project.readthedocs.io/en/latest/overview.html), [Camelot](https://camelotproject.org/), [WildePod](https://wildepod.org/), [wpsWatch](https://wildlabs.net/inventory/products/wpswatch), [TNC Animl](https://animl.camera/) ([code](https://github.com/tnc-ca-geo/animl-frontend)), [Wildlife Observer Network](https://roadecology.ucdavis.edu/research/projects/wildlife-observer-network), [Zooniverse ML Subject Assistant](https://subject-assistant.zooniverse.org/#/intro), [Dudek AI Image Toolkit](https://ait.dudek.com), [Zamba Cloud](https://github.com/drivendataorg/zamba), [OCAPI](https://www.terroiko.fr/ocapi/).
+* The [list of MD-related GUIs, platforms, and GitHub repos](https://github.com/agentmorris/MegaDetector/blob/main/megadetector.md#is-there-a-gui) within the MegaDetector User Guide... although you can never have too many lists, so here they are in a concise comma-separated list: [Wildlife Insights](https://wildlifeinsights.org/), [Animal Detect](https://www.animaldetect.com), [TrapTagger](https://wildeyeconservation.org/trap-tagger-about/), [WildTrax](https://www.wildtrax.ca/), [Agouti](https://agouti.eu/), [Trapper](https://trapper-project.readthedocs.io/en/latest/overview.html), [Camelot](https://camelotproject.org/), [WildePod](https://wildepod.org/), [wpsWatch](https://wildlabs.net/inventory/products/wpswatch), [TNC Animl](https://animl.camera/) ([code](https://github.com/tnc-ca-geo/animl-frontend)), [Wildlife Observer Network](https://roadecology.ucdavis.edu/research/projects/wildlife-observer-network), [Zooniverse ML Subject Assistant](https://subject-assistant.zooniverse.org/#/intro), [Dudek AI Image Toolkit](https://ait.dudek.com), [Zamba Cloud](https://github.com/drivendataorg/zamba), [OCAPI](https://www.terroiko.fr/ocapi/), [BoquilaHUB](https://boquila.org/hub)
 
 * [Peter's map of AddaxAI (formerly EcoAssist) users](https://addaxdatascience.com/addaxai/) (who are also MegaDetector users!)
 

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/data_management/cct_json_utils.py

@@ -305,6 +305,9 @@ class SequenceOptions:
         #: How to handle invalid datetimes: 'error' or 'none'
         self.datetime_conversion_failure_behavior = 'none'
 
+        #: Enable additional debug output
+        self.verbose = False
+
 
 #%% Functions
 
@@ -331,7 +334,9 @@ def write_object_with_serialized_datetimes(d,json_fn):
         json.dump(d,f,indent=1,default=json_serialize_datetime)
 
 
-def parse_datetimes_from_cct_image_list(images,conversion_failure_behavior='error'):
+def parse_datetimes_from_cct_image_list(images,
+                                        conversion_failure_behavior='error',
+                                        verbose=False):
     """
     Given the "images" field from a COCO camera traps dictionary, converts all
     string-formatted datetime fields to Python datetimes, making reasonable assumptions
@@ -342,6 +347,7 @@ def parse_datetimes_from_cct_image_list(images,conversion_failure_behavior='erro
         conversion_failure_behavior (str, optional): determines what happens on a failed
             conversion; can be "error" (raise an error), "str" (leave as a string), or
             "none" (convert to None)
+        verbose (bool, optional): enable additional debug output
 
     Returns:
         images: the input list, with datetimes converted (after modifying in place)
@@ -359,14 +365,17 @@ def parse_datetimes_from_cct_image_list(images,conversion_failure_behavior='erro
             dt = dateutil.parser.parse(im['datetime'])
             im['datetime'] = dt
         except Exception as e:
-            s = 'could not parse datetime {}: {}'.format(str(im['datetime']),str(e))
+            s = 'could not parse datetime {} from {}: {}'.format(
+                str(im['datetime']),im['file_name'],str(e))
             if conversion_failure_behavior == 'error':
                 raise ValueError(s)
             elif conversion_failure_behavior == 'str':
-                print('Warning: {}'.format(s))
+                if verbose:
+                    print('Warning: {}'.format(s))
                 pass
             elif conversion_failure_behavior == 'none':
-                print('Warning: {}'.format(s))
+                if verbose:
+                    print('Warning: {}'.format(s))
                 im['datetime'] = None
 
     # ...for each image
@@ -450,7 +459,8 @@ def create_sequences(image_info,options=None):
 
     # Modifies the images in place
     _ = parse_datetimes_from_cct_image_list(image_info,
-            conversion_failure_behavior=options.datetime_conversion_failure_behavior)
+            conversion_failure_behavior=options.datetime_conversion_failure_behavior,
+            verbose=options.verbose)
 
     n_invalid_datetimes = 0
     for im in image_info:
@@ -505,7 +515,7 @@ def create_sequences(image_info,options=None):
                 delta = (im['datetime'] - previous_datetime).total_seconds()
 
             # Start a new sequence if necessary, including the case where this datetime is invalid
-            if delta is None or delta > options.episode_interval_seconds or invalid_datetime:
+            if (delta is None) or (delta > options.episode_interval_seconds) or (invalid_datetime):
                 next_frame_number = 0
                 current_sequence_id = 'location_{}_sequence_index_{}'.format(
                     location,str(next_sequence_number).zfill(5))

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/data_management/databases/subset_json_db.py

@@ -18,13 +18,20 @@ import json
 import argparse
 
 from tqdm import tqdm
-from megadetector.utils import ct_utils
 from copy import copy
 
+from megadetector.utils import ct_utils
+from megadetector.utils.ct_utils import sort_list_of_dicts_by_key
+
 
 #%% Functions
 
-def subset_json_db(input_json, query, output_json=None, ignore_case=False, verbose=False):
+def subset_json_db(input_json,
+                   query,
+                   output_json=None,
+                   ignore_case=False,
+                   remap_categories=True,
+                   verbose=False):
     """
     Given a json file (or dictionary already loaded from a json file), produce a new
     database containing only the images whose filenames contain the string 'query',
@@ -36,6 +43,8 @@ def subset_json_db(input_json, query, output_json=None, ignore_case=False, verbo
             contain this string.  If this is a list, test for exact matches.
         output_json (str, optional): file to write the resulting .json file to
         ignore_case (bool, optional): whether to perform a case-insensitive search for [query]
+        remap_categories (bool, optional): trim the category list to only the categores used
+            in the subset
         verbose (bool, optional): enable additional debug output
 
     Returns:
@@ -92,6 +101,52 @@ def subset_json_db(input_json, query, output_json=None, ignore_case=False, verbo
     output_data['images'] = images
     output_data['annotations'] = annotations
 
+    # Remap categories if necessary
+    if remap_categories:
+
+        category_ids_used = set()
+        for ann in annotations:
+            category_ids_used.add(ann['category_id'])
+
+        if verbose:
+            print('Keeping {} of {} categories'.format(
+                len(category_ids_used),len(input_data['categories'])))
+
+        input_category_id_to_output_category_id = {}
+
+        next_category_id = 0
+
+        # Build mappings from old to new category IDs
+        for input_category_id in category_ids_used:
+            assert isinstance(input_category_id,int), \
+                'Illegal category ID {}'.format(input_category_id)
+            output_category_id = next_category_id
+            next_category_id = next_category_id + 1
+            input_category_id_to_output_category_id[input_category_id] = output_category_id
+
+        # Modify the annotations
+        for ann in annotations:
+            assert ann['category_id'] in input_category_id_to_output_category_id
+            ann['category_id'] = input_category_id_to_output_category_id[ann['category_id']]
+
+        output_categories = []
+
+        # Re-write the category table
+        for cat in input_data['categories']:
+
+            if cat['id'] in input_category_id_to_output_category_id:
+
+                # There may be non-required fields, so don't just create an empty dict
+                # and copy the name/id field, keep the original dict other than "id"
+                output_category = copy(cat)
+                output_category['id'] = input_category_id_to_output_category_id[cat['id']]
+                output_categories.append(output_category)
+
+        output_categories = sort_list_of_dicts_by_key(output_categories,'id')
+        output_data['categories'] = output_categories
+
+    # ...if we need to remap categories
+
     # Write the output file if requested
     if output_json is not None:
         if verbose:

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/pytorch_detector.py

@@ -328,7 +328,8 @@ def _initialize_yolo_imports(model_type='yolov5',
     if try_yolov5_import and not utils_imported:
 
         try:
-            from yolov5.utils.general import non_max_suppression, xyxy2xywh # noqa
+            # from yolov5.utils.general import non_max_suppression # type: ignore
+            from yolov5.utils.general import xyxy2xywh # noqa
             from yolov5.utils.augmentations import letterbox # noqa
             try:
                 from yolov5.utils.general import scale_boxes as scale_coords
@@ -348,7 +349,8 @@ def _initialize_yolo_imports(model_type='yolov5',
 
         try:
 
-            from yolov9.utils.general import non_max_suppression, xyxy2xywh # noqa
+            # from yolov9.utils.general import non_max_suppression # noqa
+            from yolov9.utils.general import xyxy2xywh # noqa
             from yolov9.utils.augmentations import letterbox # noqa
             from yolov9.utils.general import scale_boxes as scale_coords # noqa
             utils_imported = True
@@ -378,7 +380,12 @@ def _initialize_yolo_imports(model_type='yolov5',
 
         try:
 
-            from ultralytics.utils.ops import non_max_suppression # type: ignore # noqa
+            # The non_max_suppression() function moved from the ops module to the nms module
+            # in mid-2025
+            try:
+                from ultralytics.utils.ops import non_max_suppression # type: ignore # noqa
+            except Exception:
+                from ultralytics.utils.nms import non_max_suppression # type: ignore # noqa
             from ultralytics.utils.ops import xyxy2xywh # type: ignore # noqa
 
             # In the ultralytics package, scale_boxes and scale_coords both exist;
@@ -444,9 +451,9 @@ def _initialize_yolo_imports(model_type='yolov5',
             if verbose:
                 print('Imported utils from ultralytics package')
 
-        except Exception:
+        except Exception as e:
 
-            # print('Ultralytics module import failed')
+            print('Ultralytics module import failed: {}'.format(str(e)))
             pass
 
     # If we haven't succeeded yet, assume the YOLOv5 repo is on our PYTHONPATH.
@@ -455,7 +462,8 @@ def _initialize_yolo_imports(model_type='yolov5',
         try:
 
             # import pre- and post-processing functions from the YOLOv5 repo
-            from utils.general import non_max_suppression, xyxy2xywh # type: ignore
+            # from utils.general import non_max_suppression # type: ignore
+            from utils.general import xyxy2xywh # type: ignore
             from utils.augmentations import letterbox # type: ignore
 
             # scale_coords() is scale_boxes() in some YOLOv5 versions
@@ -851,6 +859,9 @@ class PTDetector:
             except AttributeError:
                 pass
 
+        # AddaxAI depends on this printout, don't remove it
+        print('PTDetector using device {}'.format(str(self.device).lower()))
+
         try:
             self.model = PTDetector._load_model(model_path,
                                                 device=self.device,
@@ -1283,17 +1294,23 @@ class PTDetector:
         else:
             nms_iou_thres = 0.6
 
-        pred = nms(prediction=pred,
-                   conf_thres=detection_threshold,
-                   iou_thres=nms_iou_thres)
+        use_library_nms = False
+
+        # Model output format changed in recent ultralytics packages, and the nms implementation
+        # in this module hasn't been updated to handle that format yet.
+        if (yolo_model_type_imported is not None) and (yolo_model_type_imported == 'ultralytics'):
+            use_library_nms = True
 
-        # For posterity, the ultralytics implementation
-        if False:
+        if use_library_nms:
             pred = non_max_suppression(prediction=pred,
-                                       conf_thres=detection_threshold,
-                                       iou_thres=nms_iou_thres,
-                                       agnostic=False,
-                                       multi_label=False)
+                                    conf_thres=detection_threshold,
+                                    iou_thres=nms_iou_thres,
+                                    agnostic=False,
+                                    multi_label=False)
+        else:
+            pred = nms(prediction=pred,
+                    conf_thres=detection_threshold,
+                    iou_thres=nms_iou_thres)
 
         assert isinstance(pred, list)
         assert len(pred) == len(batch_metadata), \

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/run_detector.py

@@ -596,8 +596,7 @@ def load_detector(model_file,
     model_file = try_download_known_detector(model_file,
                                              force_download=force_model_download)
 
-    if verbose:
-        print('GPU available: {}'.format(is_gpu_available(model_file)))
+    print('GPU available: {}'.format(is_gpu_available(model_file)))
 
     start_time = time.time()
 

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/run_detector_batch.py

@@ -94,20 +94,29 @@ max_queue_size = 10
 # How often should we print progress when using the image queue?
 n_queue_print = 1000
 
-# TODO: it's a little sloppy that these are module-level globals, but in practice it
+# Only used if --include_exif_data or --include_image_timestamp are supplied
+exif_options = read_exif.ReadExifOptions()
+exif_options.processing_library = 'pil'
+exif_options.byte_handling = 'convert_to_string'
+
+# Only relevant when we're running our test harness; because bugs in batch
+# inference are dependent on batch grouping, we randomize batch grouping
+# during testing to maximize the probability that latent bugs come up
+# eventually.
+randomize_batch_order_during_testing = True
+
+# TODO: it's a little sloppy that the following are module-level globals, but in practice it
 # doesn't really matter, so I'm not in a big rush to move these to options until I do
 # a larger cleanup of all the long argument lists in this module.
-#
+
 # Should the consumer loop run on its own process, or here in the main process?
 run_separate_consumer_process = False
-use_threads_for_queue = False
-verbose = False
 
-exif_options = read_exif.ReadExifOptions()
-exif_options.processing_library = 'pil'
-exif_options.byte_handling = 'convert_to_string'
+# Should we use threads (rather than processes) for the data loading workers?
+use_threads_for_queue = False
 
-randomize_batch_order_during_testing = True
+# Enable additional debug output
+verbose = False
 
 
 #%% Support functions for multiprocessing
@@ -736,7 +745,9 @@ def _process_batch(image_items_batch,
         try:
 
             batch_detections = \
-                detector.generate_detections_one_batch(valid_images, valid_image_filenames, verbose=verbose)
+                detector.generate_detections_one_batch(valid_images,
+                                                       valid_image_filenames,
+                                                       verbose=verbose)
 
             assert len(batch_detections) == len(valid_images)
 
@@ -1050,7 +1061,8 @@ def load_and_run_detector_batch(model_file,
                                 detector_options=None,
                                 loader_workers=default_loaders,
                                 preprocess_on_image_queue=default_preprocess_on_image_queue,
-                                batch_size=1):
+                                batch_size=1,
+                                verbose_output=False):
     """
     Load a model file and run it on a list of images.
 
@@ -1087,6 +1099,7 @@ def load_and_run_detector_batch(model_file,
         preprocess_on_image_queue (bool, optional): if the image queue is enabled, should it handle
             image loading and preprocessing (True), or just image loading (False)?
         batch_size (int, optional): batch size for GPU processing, automatically set to 1 for CPU processing
+        verbose_output (bool, optional): enable additional debug output
 
     Returns:
         results: list of dicts; each dict represents detections on one image
@@ -1109,6 +1122,11 @@ def load_and_run_detector_batch(model_file,
     if class_mapping_filename is not None:
         _load_custom_class_mapping(class_mapping_filename)
 
+    global verbose
+    if verbose_output:
+        print('Enabling verbose output')
+        verbose = True
+
     # Handle the case where image_file_names is not yet actually a list
     if isinstance(image_file_names,str):
 
@@ -1866,11 +1884,7 @@ def main(): # noqa
 
     args = parser.parse_args()
 
-    global verbose
     global use_threads_for_queue
-
-    if args.verbose:
-        verbose = True
     if args.use_threads_for_queue:
         use_threads_for_queue = True
 
@@ -2087,7 +2101,8 @@ def main(): # noqa
                                           detector_options=detector_options,
                                           loader_workers=args.loader_workers,
                                           preprocess_on_image_queue=args.preprocess_on_image_queue,
-                                          batch_size=args.batch_size)
+                                          batch_size=args.batch_size,
+                                          verbose_output=args.verbose)
 
     elapsed = time.time() - start_time
     images_per_second = len(results) / elapsed

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/run_inference_with_yolov5_val.py

@@ -159,7 +159,9 @@ class YoloInferenceOptions:
         #: These are deliberately offset from the standard MD categories; YOLOv5
         #: needs categories IDs to start at 0.
         #:
-        #: This can also be a string that points to a YOLO dataset.yaml file.
+        #: This can also be a string that points to any class mapping file supported
+        #: by read_classes_from_yolo_dataset_file(): a YOLO dataset.yaml file, a text
+        #: file with a list of classes, or a .json file with an ID --> name dict
         self.yolo_category_id_to_name = {0:'animal',1:'person',2:'vehicle'}
 
         #: What should we do if the output file already exists?

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/run_tiled_inference.py

@@ -39,7 +39,7 @@ from torchvision import ops
 from megadetector.detection.run_inference_with_yolov5_val import \
     YoloInferenceOptions,run_inference_with_yolo_val
 from megadetector.detection.run_detector_batch import \
-    load_and_run_detector_batch,write_results_to_file
+    load_and_run_detector_batch,write_results_to_file,default_loaders
 from megadetector.detection.run_detector import \
     try_download_known_detector, CONF_DIGITS, COORD_DIGITS
 from megadetector.utils import path_utils
@@ -406,7 +406,9 @@ def run_tiled_inference(model_file,
                         detector_options=None,
                         use_image_queue=True,
                         preprocess_on_image_queue=True,
-                        inference_size=None):
+                        loader_workers=default_loaders,
+                        inference_size=None,
+                        verbose=False):
     """
     Runs inference using [model_file] on the images in [image_folder], fist splitting each image up
     into tiles of size [tile_size_x] x [tile_size_y], writing those tiles to [tiling_folder],
@@ -451,16 +453,17 @@ def run_tiled_inference(model_file,
         image_list (list, optional): .json file containing a list of specific images to process.  If
             this is supplied, and the paths are absolute, [image_folder] will be ignored. If this is supplied,
             and the paths are relative, they should be relative to [image_folder]
-        augment (bool, optional): apply test-time augmentation, only relevant if yolo_inference_options
-            is None
+        augment (bool, optional): apply test-time augmentation
         detector_options (dict, optional): parameters to pass to run_detector, only relevant if
             yolo_inference_options is None
         use_image_queue (bool, optional): whether to use a loader worker queue, only relevant if
             yolo_inference_options is None
         preprocess_on_image_queue (bool, optional): whether the image queue should also be responsible
             for preprocessing
+        loader_workers (int, optional): number of preprocessing loader workers to use
         inference_size (int, optional): override the default inference image size, only relevant if
             yolo_inference_options is None
+        verbose (bool, optional): enable additional debug output
 
     Returns:
         dict: MD-formatted results dictionary, identical to what's written to [output_file]
@@ -522,7 +525,8 @@ def run_tiled_inference(model_file,
 
     all_image_patch_info = None
 
-    print('Extracting patches from {} images'.format(len(image_files_relative)))
+    print('Extracting patches from {} images on {} workers'.format(
+        len(image_files_relative),n_patch_extraction_workers))
 
     n_workers = n_patch_extraction_workers
 
@@ -632,7 +636,9 @@ def run_tiled_inference(model_file,
                                                         detector_options=detector_options,
                                                         use_image_queue=use_image_queue,
                                                         preprocess_on_image_queue=preprocess_on_image_queue,
-                                                        image_size=inference_size)
+                                                        image_size=inference_size,
+                                                        verbose_output=verbose,
+                                                        loader_workers=loader_workers)
 
         patch_level_output_file = os.path.join(tiling_folder,folder_name + '_patch_level_results.json')
 
@@ -847,12 +853,12 @@ if False:
         yolo_inference_options.yolo_working_folder = os.path.expanduser('~/git/yolov5')
 
     run_tiled_inference(model_file, image_folder, tiling_folder, output_file,
-                            tile_size_x=tile_size_x, tile_size_y=tile_size_y,
-                            tile_overlap=tile_overlap,
-                            checkpoint_path=checkpoint_path,
-                            checkpoint_frequency=checkpoint_frequency,
-                            remove_tiles=remove_tiles,
-                            yolo_inference_options=yolo_inference_options)
+                        tile_size_x=tile_size_x, tile_size_y=tile_size_y,
+                        tile_overlap=tile_overlap,
+                        checkpoint_path=checkpoint_path,
+                        checkpoint_frequency=checkpoint_frequency,
+                        remove_tiles=remove_tiles,
+                        yolo_inference_options=yolo_inference_options)
 
 
     #%% Run tiled inference (generate a command)
@@ -907,7 +913,10 @@ if False:
 
 #%% Command-line driver
 
-def main(): # noqa
+def main():
+    """
+    Command-line driver for run_tiled_inference
+    """
 
     parser = argparse.ArgumentParser(
         description='Chop a folder of images up into tiles, run MD on the tiles, and stitch the results together')
@@ -927,6 +936,14 @@ def main(): # noqa
         '--no_remove_tiles',
         action='store_true',
         help='Tiles are removed by default; this option suppresses tile deletion')
+    parser.add_argument(
+        '--augment',
+        action='store_true',
+        help='Enable test-time augmentation')
+    parser.add_argument(
+        '--verbose',
+        action='store_true',
+        help='Enable additional debug output')
     parser.add_argument(
         '--tile_size_x',
         type=int,
@@ -956,7 +973,22 @@ def main(): # noqa
         '--detector_options',
         type=str,
         default=None,
-        help=('A list of detector options (key-value pairs) to '))
+        help=('A list of detector options (key-value pairs)'))
+    parser.add_argument(
+        '--inference_size',
+        type=int,
+        default=None,
+        help=('Run inference at a non-default size'))
+    parser.add_argument(
+        '--n_patch_extraction_workers',
+        type=int,
+        default=1,
+        help=('Number of workers to use for patch extraction'))
+    parser.add_argument(
+        '--loader_workers',
+        type=int,
+        default=default_loaders,
+        help=('Number of workers to use for image loading and preprocessing (0 to disable)'))
 
     # detector_options = parse_kvp_list(args.detector_options)
 
@@ -984,11 +1016,23 @@ def main(): # noqa
 
     remove_tiles = (not args.no_remove_tiles)
 
-    run_tiled_inference(model_file, args.image_folder, args.tiling_folder, args.output_file,
-                        tile_size_x=args.tile_size_x, tile_size_y=args.tile_size_y,
+    use_image_queue = (args.loader_workers > 0)
+
+    run_tiled_inference(model_file,
+                        args.image_folder,
+                        args.tiling_folder,
+                        args.output_file,
+                        tile_size_x=args.tile_size_x,
+                        tile_size_y=args.tile_size_y,
                         tile_overlap=args.tile_overlap,
                         remove_tiles=remove_tiles,
-                        image_list=args.image_list)
+                        image_list=args.image_list,
+                        augment=args.augment,
+                        inference_size=args.inference_size,
+                        verbose=args.verbose,
+                        n_patch_extraction_workers=args.n_patch_extraction_workers,
+                        loader_workers=args.loader_workers,
+                        use_image_queue=use_image_queue)
 
 if __name__ == '__main__':
     main()

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/detection/video_utils.py

@@ -22,6 +22,7 @@ from functools import partial
 from inspect import signature
 
 from megadetector.utils import path_utils
+from megadetector.utils.path_utils import clean_path
 from megadetector.utils.ct_utils import sort_list_of_dicts_by_key
 from megadetector.visualization import visualization_utils as vis_utils
 
@@ -592,7 +593,7 @@ def video_to_frames(input_video_file,
                     quality=None,
                     max_width=None,
                     frames_to_extract=None,
-                    allow_empty_videos=False):
+                    allow_empty_videos=True):
     """
     Renders frames from [input_video_file] to .jpg files in [output_folder].
 
@@ -618,8 +619,8 @@ def video_to_frames(input_video_file,
             a single frame number.  In the special case where frames_to_extract
             is [], this function still reads video frame rates and verifies that videos
             are readable, but no frames are extracted.
-        allow_empty_videos (bool, optional): Just print a warning if a video appears to have no
-            frames (by default, this is an error).
+        allow_empty_videos (bool, optional): Just print a warning if a video appears to have
+            no frames (by default, this is an error).
 
     Returns:
         tuple: length-2 tuple containing (list of frame filenames,frame rate)
@@ -883,7 +884,14 @@ def _video_to_frames_for_folder(relative_fn,input_folder,output_folder_base,
 
     # Create the target output folder
     output_folder_video = os.path.join(output_folder_base,relative_fn)
-    os.makedirs(output_folder_video,exist_ok=True)
+    try:
+        os.makedirs(output_folder_video,exist_ok=True)
+    except Exception:
+        output_folder_clean = clean_path(output_folder_video)
+        print('Warning: failed to create folder {}, trying {}'.format(
+            output_folder_video,output_folder_clean))
+        output_folder_video = output_folder_clean
+        os.makedirs(output_folder_video,exist_ok=True)
 
     # Render frames
     # input_video_file = input_fn_absolute; output_folder = output_folder_video
@@ -1090,6 +1098,9 @@ class FrameToVideoOptions:
         #: Are frame rates required?
         self.frame_rates_are_required = False
 
+        #: Enable additional debug output
+        self.verbose = False
+
 
 def frame_results_to_video_results(input_file,
                                    output_file,
@@ -1176,9 +1187,14 @@ def frame_results_to_video_results(input_file,
 
         if (video_filename_to_frame_rate is not None):
 
-            if options.frame_rates_are_required:
-                assert video_name in video_filename_to_frame_rate, \
-                    'Could not determine frame rate for {}'.format(video_name)
+            if video_name not in video_filename_to_frame_rate:
+
+                s = 'Could not determine frame rate for {}'.format(video_name)
+                if options.frame_rates_are_required:
+                    raise ValueError(s)
+                elif options.verbose:
+                    print('Warning: {}'.format(s))
+
             if video_name in video_filename_to_frame_rate:
                 im_out['frame_rate'] = video_filename_to_frame_rate[video_name]
 

{megadetector-10.0.7 → megadetector-10.0.9}/megadetector/postprocessing/classification_postprocessing.py

@@ -130,7 +130,7 @@ class ClassificationSmoothingOptions:
 
         ## Populated internally
 
-        #: #: Only include these categories in the smoothing process (None to use all categories)
+        #: Only include these categories in the smoothing process (None to use all categories)
         self._detection_category_ids_to_smooth = None
 
 
@@ -1015,6 +1015,10 @@ def smooth_classification_results_sequence_level(input_file,
 
         detections_this_sequence = []
         for image_filename in image_filenames_this_sequence:
+            if image_filename not in image_fn_to_classification_results:
+                print('Warning: {} in sequence list but not in results'.format(
+                    image_filename))
+                continue
             im = image_fn_to_classification_results[image_filename]
             if 'detections' not in im or im['detections'] is None:
                 continue