Add prefix mmdet to LoadImageFromNDArray

[Danil328]

Motivation

I use mmdeploy in my project.

Modification

There was an error when using a model from the mmyolo repository during image preprocessing, provided that an np.ndarray was supplied as input. If you input the path to the image, everything works fine.

Error reproduction code:

from mmdeploy.apis.utils import build_task_processor
from mmdeploy.utils import get_input_shape, load_config
import torch
import cv2

deploy_cfg = './configs/deploy/detection_tensorrt_dynamic-192x192-960x960.py'
model_cfg = '../mmyolo/configs/yolov5/yolov5_s-v61_syncbn_8xb16-300e_coco.py'
device = 'cuda:0'
backend_model = ['./work_dir/end2end.engine']
image_str = '../mmyolo/demo/demo.jpg'
image_np = cv2.imread(image_str)

# read deploy_cfg and model_cfg
deploy_cfg, model_cfg = load_config(deploy_cfg, model_cfg)

# build task and backend model
task_processor = build_task_processor(model_cfg, deploy_cfg, device)
model = task_processor.build_backend_model(backend_model)

# process input image
input_shape = get_input_shape(deploy_cfg)
model_inputs, _ = task_processor.create_input(image_str, input_shape)  # Success
model_inputs, _ = task_processor.create_input(image_np, input_shape)  # Fails

Error:

KeyError: 'LoadImageFromNDArray is not in the transform registry. Please check whether the value of `LoadImageFromNDArray` is correct or it was registered as expected.

Also I found similar issue - #2197

[CLAassistant]

All committers have signed the CLA.

[Danil328]

I made minor changes, and I see that the tests are failing for a completely different reason, what should I do in this case?

[RunningLeon]

LGTM