提交 09e72110 编写于 作者: L LDOUBLEV

fix bugs and adjust warmup times from 10 to 2 to speed up

上级 e450e32d
......@@ -138,8 +138,8 @@ infer_img_dir=$(func_parser_value "${lines[44]}")
save_log_key=$(func_parser_key "${lines[45]}")
benchmark_key=$(func_parser_key "${lines[46]}")
benchmark_value=$(func_parser_value "${lines[46]}")
infer_key2=$(func_parser_key "${lines[47]}")
infer_value2=$(func_parser_value "${lines[47]}")
infer_key1=$(func_parser_key "${lines[47]}")
infer_value1=$(func_parser_value "${lines[47]}")
LOG_PATH="./tests/output"
mkdir -p ${LOG_PATH}
......@@ -169,7 +169,8 @@ function func_inference(){
set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
set_cpu_threads=$(func_set_params "${cpu_threads_key}" "${threads}")
set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} > ${_save_log_path} 2>&1 "
set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} > ${_save_log_path} 2>&1 "
eval $command
status_check $? "${command}" "${status_log}"
done
......@@ -285,7 +286,8 @@ for gpu in ${gpu_list[*]}; do
set_eval_pretrain=$(func_set_params "${pretrain_model_key}" "${save_log}/${train_model_name}")
# run eval
if [ ${eval_py} != "null" ]; then
eval_cmd="${python} ${eval_py} ${set_eval_pretrain} ${set_use_gpu}"
set_eval_params1=$(func_set_params "${eval_key1}" "${eval_value1}")
eval_cmd="${python} ${eval_py} ${set_eval_pretrain} ${set_use_gpu} ${set_eval_params1}"
eval $eval_cmd
status_check $? "${eval_cmd}" "${status_log}"
fi
......@@ -318,3 +320,4 @@ else
#run inference
func_inference "${python}" "${inference_py}" "${infer_model}" "${LOG_PATH}" "${infer_img_dir}" "False"
fi
......@@ -114,7 +114,7 @@ class TextDetector(object):
time_keys=[
'preprocess_time', 'inference_time', 'postprocess_time'
],
warmup=10,
warmup=2,
logger=logger)
def order_points_clockwise(self, pts):
......@@ -237,7 +237,7 @@ if __name__ == "__main__":
if args.warmup:
img = np.random.uniform(0, 255, [640, 640, 3]).astype(np.uint8)
for i in range(10):
for i in range(2):
res = text_detector(img)
if not os.path.exists(draw_img_save):
......
......@@ -73,7 +73,7 @@ class TextRecognizer(object):
model_precision=args.precision,
batch_size=args.rec_batch_num,
data_shape="dynamic",
save_path=args.save_log_path,
save_path=None, #args.save_log_path,
inference_config=self.config,
pids=pid,
process_name=None,
......@@ -81,7 +81,8 @@ class TextRecognizer(object):
time_keys=[
'preprocess_time', 'inference_time', 'postprocess_time'
],
warmup=10)
warmup=2,
logger=logger)
def resize_norm_img(self, img, max_wh_ratio):
imgC, imgH, imgW = self.rec_image_shape
......@@ -272,10 +273,10 @@ def main(args):
valid_image_file_list = []
img_list = []
# warmup 10 times
# warmup 2 times
if args.warmup:
img = np.random.uniform(0, 255, [32, 320, 3]).astype(np.uint8)
for i in range(10):
for i in range(2):
res = text_recognizer([img])
for image_file in image_file_list:
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册