[CodeCamp2023-500]add large_image_demo

[YanxingLiu]

Motivation

To add large image demo in mmdetection.

Modification

demo/large_image.jpg
demo/large_image_demo.py
mmdet/utils/large_image.py
mmdet/utils/misc.py

Use cases (Optional)

wget -P checkpoint https://download.openmmlab.com/mmdetection/v2.0/faster_rcnn/faster_rcnn_r101_fpn_2x_coco/faster_rcnn_r101_fpn_2x_coco_bbox_mAP-0.398_20200504_210455-1d2dac9c.pth # noqa: E501, E261.

python demo/large_image_demo.py \
    demo/large_image.jpg \
    configs/faster_rcnn/faster-rcnn_r101_fpn_2x_coco.py \
    checkpoint/faster_rcnn_r101_fpn_2x_coco_bbox_mAP-0.398_20200504_210455-1d2dac9c.pth

result:

parameter test:

• cpu inference: pass
• --show: pass
  
• --tta: pass

wget -P checkpoint https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_r50_fpn_1x_coco/retinanet_r50_fpn_1x_coco_20200130-c2398f9e.pth # noqa: E501, E261.

python demo/large_image_demo.py \
    demo/large_image.jpg \
    configs/retinanet/retinanet_r50_fpn_1x_coco.py \
    checkpoint/retinanet_r50_fpn_1x_coco_20200130-c2398f9e.pth --tta

result:

• --score-thr: pass
• --path-size: pass
• --patch-overlap-ratio: pass
• --merge-iou-thr: pass
• --merge-nms-type: pass
• --batch-size: pass
• --debug: pass
• --save-patch: pass

[CLAassistant]

All committers have signed the CLA.

[hhaAndroid]

@YanxingLiu Is it possible to support instance segmentation?

[YanxingLiu]

@YanxingLiu Is it possible to support instance segmentation?

I am not familiar with instance segmentation, but it feels like it might be more difficult than object detection.

[hhaAndroid]

@YanxingLiu Is it possible to support instance segmentation?

I am not familiar with instance segmentation, but it feels like it might be more difficult than object detection.

OK

[W-hary]

When I run 'large_image_demo.py' for object detection on 'large image.jpg', I can successfully get the detection results. But when I use 'rtmdet-ins_tiny_8xb32-300e_coco.py' to instance split 'large image.jpg', an error will be reported: AssertionError: binary_masks must have the same shape with image
Does 'large_image_demo.py' not work for instance split testing? How can I solve this problem

[DAIJinJ]

When I run 'large_image_demo.py' for object detection on 'large image.jpg', I can successfully get the detection results. But when I use 'rtmdet-ins_tiny_8xb32-300e_coco.py' to instance split 'large image.jpg', an error will be reported: AssertionError: binary_masks must have the same shape with image Does 'large_image_demo.py' not work for instance split testing? How can I solve this problem

same problem