python source code of bbox

Project: Collaborative-Learning-for-Weakly-Supervised-Object-Detection (GitHub Link)

Collaborative-Learning-for-Weakly-Supervised-Object-Detection-master
- LICENSE
- lib
  - datasets
    - coco.py
    - imdb.py
    - VOCdevkit-matlab-wrapper
      - xVOCap.m
      - voc_eval.m
      - get_voc_opts.m
    - __pycache__
      - pascal_voc.cpython-36.pyc
      - ds_utils.cpython-36.pyc
      - factory.cpython-36.pyc
      - __init__.cpython-36.pyc
      - dis_eval.cpython-36.pyc
      - voc_eval.cpython-36.pyc
      - imdb.cpython-36.pyc
      - coco.cpython-36.pyc
    - voc_eval.py
    - factory.py
    - dis_eval.py
    - __init__.py
    - pascal_voc.py
    - tools
      - mcg_munge.py
    - ds_utils.py
    - dis_eval.pyc
  - utils
    - blob.py
    - __pycache__
      - bbox.cpython-36.pyc
      - blob.cpython-36.pyc
      - timer.cpython-36.pyc
      - visualization.cpython-36.pyc
      - __init__.cpython-36.pyc
    - bbox.py
    - __init__.py
    - visualization.py
    - .gitignore
    - timer.py
  - roi_data_layer
    - layer.py
    - __pycache__
      - roidb.cpython-36.pyc
      - layer.cpython-36.pyc
      - __init__.cpython-36.pyc
      - minibatch.cpython-36.pyc
    - roidb.py
    - __init__.py
    - minibatch.py
  - nms
    - build.py
    - src
      - cuda
        nms_kernel.cu.o
        nms_kernel.cu
        nms_kernel.h
      - nms.c
      - nms_cuda.h
      - nms.h
      - nms_cuda.c
    - pth_nms.py
    - __init__.py
  - layer_utils
    - proposal_target_layer.py
    - generate_pseudo_gtbox.py
    - proposal_layer.py
    - snippets.py
    - roi_pooling
      - build.py
      - src
        roi_pooling_cuda.c
        cuda
        roi_pooling_kernel.h
        roi_pooling_kernel.cu.o
        roi_pooling_kernel.cu
        roi_pooling.c
        roi_pooling_cuda.h
        roi_pooling.h
      - roi_pool.py
      - __init__.py
      - roi_pool_py.py
    - anchor_target_layer.py
    - generate_anchors.py
    - __init__.py
    - choose_pseudo_gt.py
    - loss_function.py
    - proposal_top_layer.py
  - nets
    - __pycache__
      - vgg16.cpython-36.pyc
      - __init__.cpython-36.pyc
      - network.cpython-36.pyc
      - resnet_v1.cpython-36.pyc
      - mobilenet_v1.cpython-36.pyc
    - mobilenet_v1.py
    - resnet_v1.py
    - __init__.py
    - network.py
    - vgg16.py
  - model
    - train_val.pyc
    - __pycache__
      - test_train.cpython-36.pyc
      - test.cpython-36.pyc
      - bbox_transform.cpython-36.pyc
      - __init__.cpython-36.pyc
      - nms_wrapper.cpython-36.pyc
      - train_val.cpython-36.pyc
      - config.cpython-36.pyc
    - config.py
    - test.pyc
    - __init__.pyc
    - train_val.py
    - config.pyc
    - __init__.py
    - test.py
    - nms_wrapper.py
    - test_train.py
    - bbox_transform.py
- experiments
  - scripts
    - train.sh
    - convert_vgg16.sh
    - test.sh
  - cfgs
    - res101.yml
    - res50.yml
    - vgg16.yml
    - mobile.yml
    - res101-lg.yml
- data
  - scripts
    - fetch_faster_rcnn_models.sh
- README.md
- tools
  - _init_paths.pyc
  - _init_paths.py
  - trainval_net.py
  - test_net.py
  - reval_discovery.py
  - demo.py
  - reval.py
  - convert_from_tensorflow_mobile.py
  - show_boxes_results.py
  - convert_from_tensorflow_vgg.py

import torch
import numpy as np

def bbox_overlaps(boxes, query_boxes):
    """
    Parameters
    ----------
    boxes: (N, 4) ndarray or tensor or variable
    query_boxes: (K, 4) ndarray or tensor or variable
    Returns
    -------
    overlaps: (N, K) overlap between boxes and query_boxes
    """
    if isinstance(boxes, np.ndarray):
        boxes = torch.from_numpy(boxes)
        query_boxes = torch.from_numpy(query_boxes)
        out_fn = lambda x: x.numpy() # If input is ndarray, turn the overlaps back to ndarray when return
    else:
        out_fn = lambda x: x

    box_areas = (boxes[:, 2] - boxes[:, 0] + 1) * \
            (boxes[:, 3] - boxes[:, 1] + 1)
    query_areas = (query_boxes[:, 2] - query_boxes[:, 0] + 1) * \
            (query_boxes[:, 3] - query_boxes[:, 1] + 1)

    iw = (torch.min(boxes[:, 2:3], query_boxes[:, 2:3].t()) - torch.max(boxes[:, 0:1], query_boxes[:, 0:1].t()) + 1).clamp(min=0)
    ih = (torch.min(boxes[:, 3:4], query_boxes[:, 3:4].t()) - torch.max(boxes[:, 1:2], query_boxes[:, 1:2].t()) + 1).clamp(min=0)
    ua = box_areas.view(-1, 1) + query_areas.view(1, -1) - iw * ih
    overlaps = iw * ih / ua
    return out_fn(overlaps)