def populate_inference_requests_queue(in_project, inference_processes,
request_queue):
for in_dataset in in_project:
for in_item_name in in_dataset:
logger.trace('Will process image',
extra={
'dataset_name': in_dataset.name,
'image_name': in_item_name
})
in_item_paths = in_dataset.get_item_paths(in_item_name)
req = InferenceRequest(ds_name=in_dataset.name,
item_name=in_item_name,
item_paths=in_item_paths)
while True:
try:
# Set a finite timeout for the queue insertion attempt to prevent deadlocks if all of the
# inference processes die in the interim.
request_queue.put(req, timeout=0.1)
break
except queue.Full:
# If any of the inference processes has died, stop populating the requests queue and exit right
# away. Otherwise a deadlock is possible if no inference processes survive to take requests off
# the queue.
if not all(p.is_alive() for p in inference_processes):
# Early exit, return False to indicate failure.
return False
return True
def run_inference(self):
inference_mode = InferenceModeFactory.create(
self._inference_mode_config, self._in_project.meta,
self._single_image_inference)
out_project = Project(
os.path.join(TaskPaths.RESULTS_DIR, self._in_project.name),
OpenMode.CREATE)
out_project.set_meta(inference_mode.out_meta)
progress_bar = Progress('Model applying: ',
self._in_project.total_items)
for in_dataset in self._in_project:
out_dataset = out_project.create_dataset(in_dataset.name)
for in_item_name in in_dataset:
# Use output project meta so that we get an annotation that is already in the context of the output
# project (with added object classes etc).
in_item_paths = in_dataset.get_item_paths(in_item_name)
in_img = sly_image.read(in_item_paths.img_path)
in_ann = Annotation.load_json_file(in_item_paths.ann_path,
inference_mode.out_meta)
logger.trace('Will process image',
extra={
'dataset_name': in_dataset.name,
'image_name': in_item_name
})
inference_annotation = inference_mode.infer_annotate(
in_img, in_ann)
out_dataset.add_item_file(in_item_name,
in_item_paths.img_path,
ann=inference_annotation)
progress_bar.iter_done_report()
report_inference_finished()
def populate_inference_requests_queue(in_project, inference_processes,
request_queue):
for in_dataset in in_project:
for in_item_name in in_dataset:
while True:
# If any of the inference processes has died, stop populating the requests queue and exit right away.
# Otherwise a deadlock is possible if no inference processes survive to take requests off the queue.
if all(p.is_alive() for p in inference_processes):
# Do not try to add requests to a full queue to prevent deadlocks if all of the inference processes
# die in the interim.
if not request_queue.full():
logger.trace('Will process image',
extra={
'dataset_name': in_dataset.name,
'image_name': in_item_name
})
in_item_paths = in_dataset.get_item_paths(in_item_name)
req = InferenceRequest(ds_name=in_dataset.name,
item_name=in_item_name,
item_paths=in_item_paths)
request_queue.put(req)
break # out of (while True).
else:
time.sleep(0.1)
else:
# Early exit, return False to indicate failure.
return False
return True
def _process_sample(self, x, target, output):
timg = TableImg((x.shape[1], x.shape[0]), (2, 2))
bounds = np.min(x), np.max(x)
x = (x - bounds[0]) / (bounds[1] - bounds[0]) * 255
x = cv2.cvtColor(x, cv2.COLOR_RGB2BGR)
timg.paste_item(x, 0, 0)
timg.paste_item(x, 0, 1)
sq_target = np.clip(target * self.target_multi, 0, 255)
timg.paste_item(sq_target, 1, 0)
sq_output = np.expand_dims(np.argmax(output, axis=2), axis=2)
sq_output = np.clip(sq_output * self.target_multi, 0, 255)
timg.paste_item(sq_output, 1, 1)
out_img = timg.img
if self.out_scale != 1:
out_img = sly.image.resize_inter_nearest(out_img,
frow=self.out_scale,
fcol=self.out_scale)
ofpath = os.path.join(self.odir, '{:08}.png'.format(self._next_idx))
cv2.imwrite(ofpath, out_img)
self._next_idx += 1
logger.trace('Saved debug patch: {}'.format(ofpath))
def run_inference(self):
out_project_fs = copy(self.in_project_fs)
out_project_fs.root_path = self.helper.paths.results_dir
out_project_fs.make_dirs()
out_pr_meta = self.inf_feeder.out_meta
out_pr_meta.to_dir(out_project_fs.project_path)
ia_cnt = out_project_fs.pr_structure.image_cnt
progress = sly.progress_counter_inference(cnt_imgs=ia_cnt)
for sample in self.in_project_fs:
logger.trace('Will process image',
extra={'dataset_name': sample.ds_name, 'image_name': sample.image_name})
ann_packed = sly.json_load(sample.ann_path)
ann = sly.Annotation.from_packed(ann_packed, self.helper.in_project_meta)
img = cv2.imread(sample.img_path)[:, :, ::-1]
res_ann = self.inf_feeder.feed(img, ann, self._infer_on_img_legacy)
out_ann_fpath = out_project_fs.ann_path(sample.ds_name, sample.image_name)
res_ann_packed = res_ann.pack()
sly.json_dump(res_ann_packed, out_ann_fpath)
if self.debug_copy_images:
out_img_fpath = out_project_fs.img_path(sample.ds_name, sample.image_name)
sly.ensure_base_path(out_img_fpath)
shutil.copy(sample.img_path, out_img_fpath)
progress.iter_done_report()
sly.report_inference_finished()
def main():
sly.task_verification(check_in_graph)
logger.info('DTL started')
helper = sly.DtlHelper()
net = Net(helper.graph, helper.in_project_metas, helper.paths.results_dir)
helper.save_res_meta(net.get_result_project_meta())
# is_archive = net.is_archive()
results_counter = 0
for pr_name, pr_dir in helper.in_project_dirs.items():
root_path, project_name = sly.ProjectFS.split_dir_project(pr_dir)
project_fs = sly.ProjectFS.from_disk(root_path, project_name, by_annotations=True)
progress = sly.progress_counter_dtl(pr_name, project_fs.image_cnt)
for sample in project_fs:
try:
img_desc = sly.ImageDescriptor(sample)
ann = sly.json_load(sample.ann_path)
data_el = (img_desc, ann)
export_output_generator = net.start(data_el)
for res_export in export_output_generator:
logger.trace("image processed", extra={'img_name': res_export[0][0].get_img_name()})
results_counter += 1
except Exception:
ex = {
'project_name': sample.project_name,
'ds_name': sample.ds_name,
'image_name': sample.image_name
}
logger.warn('Image was skipped because some error occured', exc_info=True, extra=ex)
progress.iter_done_report()
logger.info('DTL finished', extra={'event_type': EventType.DTL_APPLIED, 'new_proj_size': results_counter})
