mirror of
https://github.com/azaion/detections.git
synced 2026-04-22 14:36:33 +00:00
[AZ-180] Refactor detection event handling and improve SSE support
- Updated the detection image endpoint to require a channel ID for event streaming. - Introduced a new endpoint for streaming detection events, allowing clients to receive real-time updates. - Enhanced the internal buffering mechanism for detection events to manage multiple channels. - Refactored the inference module to support the new event handling structure. Made-with: Cursor
This commit is contained in:
Oleksandr Bezdieniezhnykh
+22
-55
@@ -1,6 +1,4 @@
|
||||
import io
|
||||
import os
|
||||
import tempfile
|
||||
import threading
|
||||
|
||||
import av
|
||||
@@ -14,7 +12,7 @@ from ai_config cimport AIRecognitionConfig
|
||||
from engines.inference_engine cimport InferenceEngine
|
||||
from loader_http_client cimport LoaderHttpClient
|
||||
from threading import Thread
|
||||
from engines import EngineClass
|
||||
from engines import engine_factory
|
||||
|
||||
|
||||
def ai_config_from_dict(dict data):
|
||||
@@ -76,29 +74,23 @@ cdef class Inference:
|
||||
raise Exception(res.err)
|
||||
return <bytes>res.data
|
||||
|
||||
cdef convert_and_upload_model(self, bytes source_bytes, str engine_filename, str calib_cache_path):
|
||||
cdef convert_and_upload_model(self, bytes source_bytes, str models_dir):
|
||||
try:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.CONVERTING)
|
||||
models_dir = constants_inf.MODELS_FOLDER
|
||||
model_bytes = EngineClass.convert_from_source(source_bytes, calib_cache_path)
|
||||
engine_bytes, engine_filename = engine_factory.build_from_source(source_bytes, self.loader_client, models_dir)
|
||||
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.UPLOADING)
|
||||
res = self.loader_client.upload_big_small_resource(model_bytes, engine_filename, models_dir)
|
||||
res = self.loader_client.upload_big_small_resource(engine_bytes, engine_filename, models_dir)
|
||||
if res.err is not None:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.WARNING, <str>f"Failed to upload converted model: {res.err}")
|
||||
|
||||
self._converted_model_bytes = model_bytes
|
||||
self._converted_model_bytes = engine_bytes
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ENABLED)
|
||||
except Exception as e:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ERROR, <str> str(e))
|
||||
self._converted_model_bytes = <bytes>None
|
||||
finally:
|
||||
self.is_building_engine = <bint>False
|
||||
if calib_cache_path is not None:
|
||||
try:
|
||||
os.unlink(calib_cache_path)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
cdef init_ai(self):
|
||||
constants_inf.log(<str> 'init AI...')
|
||||
@@ -110,7 +102,7 @@ cdef class Inference:
|
||||
|
||||
if self._converted_model_bytes is not None:
|
||||
try:
|
||||
self.engine = EngineClass(self._converted_model_bytes)
|
||||
self.engine = engine_factory.create(self._converted_model_bytes)
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ENABLED)
|
||||
except Exception as e:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ERROR, <str> str(e))
|
||||
@@ -119,58 +111,33 @@ cdef class Inference:
|
||||
return
|
||||
|
||||
models_dir = constants_inf.MODELS_FOLDER
|
||||
engine_filename_fp16 = EngineClass.get_engine_filename()
|
||||
if engine_filename_fp16 is not None:
|
||||
engine_filename_int8 = EngineClass.get_engine_filename(<str>"int8")
|
||||
for candidate in [engine_filename_int8, engine_filename_fp16]:
|
||||
try:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.DOWNLOADING)
|
||||
res = self.loader_client.load_big_small_resource(candidate, models_dir)
|
||||
if res.err is not None:
|
||||
raise Exception(res.err)
|
||||
self.engine = EngineClass(res.data)
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ENABLED)
|
||||
return
|
||||
except Exception:
|
||||
pass
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.DOWNLOADING)
|
||||
engine = engine_factory.load_engine(self.loader_client, models_dir)
|
||||
if engine is not None:
|
||||
self.engine = engine
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ENABLED)
|
||||
return
|
||||
|
||||
source_filename = EngineClass.get_source_filename()
|
||||
if source_filename is None:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ERROR, <str>"Pre-built engine not found and no source available")
|
||||
return
|
||||
source_filename = engine_factory.get_source_filename()
|
||||
if source_filename is None:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ERROR, <str>"No engine available and no source to build from")
|
||||
return
|
||||
|
||||
source_bytes = self.download_model(source_filename)
|
||||
|
||||
if engine_factory.has_build_step:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.WARNING, <str>"Cached engine not found, converting from source")
|
||||
source_bytes = self.download_model(source_filename)
|
||||
calib_cache_path = self._try_download_calib_cache(models_dir)
|
||||
target_engine_filename = EngineClass.get_engine_filename(<str>"int8") if calib_cache_path is not None else engine_filename_fp16
|
||||
self.is_building_engine = <bint>True
|
||||
|
||||
thread = Thread(target=self.convert_and_upload_model, args=(source_bytes, target_engine_filename, calib_cache_path))
|
||||
thread = Thread(target=self.convert_and_upload_model, args=(source_bytes, models_dir))
|
||||
thread.daemon = True
|
||||
thread.start()
|
||||
return
|
||||
else:
|
||||
self.engine = EngineClass(<bytes>self.download_model(constants_inf.AI_ONNX_MODEL_FILE))
|
||||
self.engine = engine_factory.create(source_bytes)
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ENABLED)
|
||||
self.is_building_engine = <bint>False
|
||||
except Exception as e:
|
||||
self.ai_availability_status.set_status(AIAvailabilityEnum.ERROR, <str>str(e))
|
||||
self.is_building_engine = <bint>False
|
||||
|
||||
cdef str _try_download_calib_cache(self, str models_dir):
|
||||
try:
|
||||
res = self.loader_client.load_big_small_resource(constants_inf.INT8_CALIB_CACHE_FILE, models_dir)
|
||||
if res.err is not None:
|
||||
constants_inf.log(<str>f"INT8 calibration cache not available: {res.err}")
|
||||
return <str>None
|
||||
fd, path = tempfile.mkstemp(suffix='.cache')
|
||||
with os.fdopen(fd, 'wb') as f:
|
||||
f.write(res.data)
|
||||
constants_inf.log(<str>'INT8 calibration cache downloaded')
|
||||
return <str>path
|
||||
except Exception as e:
|
||||
constants_inf.log(<str>f"INT8 calibration cache download failed: {str(e)}")
|
||||
return <str>None
|
||||
|
||||
cpdef run_detect_image(self, bytes image_bytes, AIRecognitionConfig ai_config, str media_name,
|
||||
object annotation_callback, object status_callback=None):
|
||||
cdef list all_frame_data = []
|
||||
|
||||
Reference in New Issue
Block a user