Merge load / resize / cache to optimize data loading efficiency for detection & instance segmentation

[goodsong81]

Summary

[Changes]

• Extend Merge load / resize / cache to optimize data loading efficiency for classification #2438 to detection & instance segmentation models
  • ATSS-MV2
  • SSD-MV2
  • YOLOX-Tiny
  • MRCNN-R50
  • MRCNN-EFF
  • [!] Will be applied to "Other" models in future
• Enable caching of non-uchar images

[Results (sampled)]

• peanut (1920x1080) / YOLOX (640x640) / default setting (num_workers=2)

Branch	Size	Cache	Data time	Iter time	E2E time	Accuracy
develop	256x256	off	0.3891	0.4666	94.97	1.0
	256x256	168M/2G	0.0767	0.1570	41.31	1.0
feat/load-resize-cache-detiseg	256x256	off	0.3072	0.3847	78.56	1.0
	256x256	64M/2G	0.0253	0.1001	32.05	1.0

How to test

[Example CMD]

otx train src/otx/algorithms/detection/configs/detection/cspdarknet_yolox/template.yaml --train-data-roots ./data/peanut --workspace outputs/load-resize-det/feat/load-resize-cache-detiseg/peanut_cspdarknet_yolox_256x256_cache2g params --learning_parameters.input_size 256x256 --algo_backend.mem_cache_size 2000000000

Checklist

• I have added unit tests to cover my changes.​
• I have added integration tests to cover my changes.​
• I have added e2e tests for validation.
• I have added the description of my changes into CHANGELOG in my target branch (e.g., CHANGELOG in develop).​
• I have updated the documentation in my target branch accordingly (e.g., documentation in develop).
• I have linked related issues.

License

• I submit my code changes under the same Apache License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below).

# Copyright (C) 2023 Intel Corporation
# SPDX-License-Identifier: Apache-2.0

[eunwoosh]

LGTM. I have a minor question. Please take a look.

[jaegukhyun]

Is there any reason that this changes does not affect to all data pipeline.py for example detr_data_pipeline.py