mmdetection_inference.py
# -*- coding:utf-8 -*-
import os
import argparse
import cv2
import mmcv
import numpy as np
from pycocotools.coco import COCO, maskUtils
from mmdet.apis import init_detector, inference_detector
from temp_coco_utils import coco_eval
os.environ['CUDA_VISIBLE_DEVICES'] = "1"
'''
用于生成可视化时候不同类别的颜色
'''
import colorsys
import random
def get_n_hls_colors(num):
hls_colors = []
i = 0
step = 360.0 / num
while i < 360:
h = i
s = 90 + random.random() * 10
l = 50 + random.random() * 10
_hlsc = [h / 360.0, l / 100.0, s / 100.0]
hls_colors.append(_hlsc)
i += step
return hls_colors
def ncolors(num):
rgb_colors = []
if num < 1:
return rgb_colors
hls_colors = get_n_hls_colors(num)
for hlsc in hls_colors:
_r, _g, _b = colorsys.hls_to_rgb(hlsc[0], hlsc[1], hlsc[2])
r, g, b = [int(x * 255.0) for x in (_r, _g, _b)]
rgb_colors.append([r, g, b])
return rgb_colors
'''
用于将结果保存到json文件
'''
def xyxy2xywh(bbox):
_bbox = bbox.tolist()
return [
_bbox[0],
_bbox[1],
_bbox[2] - _bbox[0] + 1,
_bbox[3] - _bbox[1] + 1,
]
def proposal2json(coco, results):
imgIds = coco.getImgIds(catIds=[])
categories = coco.dataset['categories']
json_results = []
for idx in range(len(imgIds)):
img_id = imgIds[idx]
bboxes = results[idx]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = 1
json_results.append(data)
return json_results
def det2json(coco, results):
imgIds = coco.getImgIds(catIds=[])
categories = coco.dataset['categories']
json_results = []
for idx in range(len(imgIds)):
img_id = imgIds[idx]
result = results[idx]
for label in range(len(result)):
bboxes = result[label]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = categories[label]['id']
json_results.append(data)
return json_results
def segm2json(coco, results):
imgIds = coco.getImgIds(catIds=[])
categories = coco.dataset['categories']
bbox_json_results = []
segm_json_results = []
for idx in range(len(imgIds)):
img_id = imgIds[idx]
det, seg = results[idx]
for label in range(len(det)):
# bbox results
bboxes = det[label]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = categories[label]['id']
bbox_json_results.append(data)
# segm results
# some detectors use different score for det and segm
if len(seg) == 2:
segms = seg[0][label]
mask_score = seg[1][label]
else:
segms = seg[label]
mask_score = [bbox[4] for bbox in bboxes]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['score'] = float(mask_score[i])
data['category_id'] = categories[label]['id']
segms[i]['counts'] = segms[i]['counts'].decode()
data['segmentation'] = segms[i]
segm_json_results.append(data)
return bbox_json_results, segm_json_results
'''
路径创建函数
'''
def mkdir_os(path):
if not os.path.exists(path):
os.makedirs(path)
def main(args):
eval_types = args.eval
mkdir_os(args.output_vis_result)
#异常为input图像没有GT-object,是一张纯背景图
mkdir_os(os.path.join(args.output_vis_result, "abnormalcase"))
mkdir_os(os.path.join(args.output_vis_result, "case"))
mkdir_os(os.path.join(args.output_vis_result, "badcase"))
score_thr = 0.3
model = init_detector(args.input_config_file, args.input_checkpoint_file, device='cuda:0')
model.eval()
'''
生成可视化类别颜色,colorbar-第一个颜色是backgroud,保留项不做使用
'''
cnum = 8
self_color = ncolors(cnum)
colorbar_vis = np.zeros((cnum * 30, 100, 3), dtype=np.uint8)
for ind, colo in enumerate(self_color):
k_tm = np.ones((30, 100, 3), dtype=np.uint8) * np.array([colo[-1], colo[-2], colo[-3]])
colorbar_vis[ind * 30:(ind + 1) * 30, 0:100] = k_tm
cv2.imwrite('../colorbar_vis.png', colorbar_vis)
coco = COCO(args.input_test_json)
imgIds = coco.getImgIds(catIds=[])
categories = coco.dataset['categories']
results = []
vis_imgpath = []
vis_imgid = []
vis_file_name = []
num = 0
count = len(imgIds)
for idx in range(len(imgIds)):
print(num,'/',count)
num += 1
img_id = imgIds[idx]
img_info = coco.loadImgs(img_id)[0]
file_name = img_info['file_name']
img_path = os.path.join(args.input_test_img_path, file_name)
result = inference_detector(model, img_path)
results.append(result)
vis_imgpath.append(img_path)
vis_imgid.append(img_info['id'])
vis_file_name.append(file_name)
if eval_types:
print('Starting evaluate {}'.format(' and '.join(eval_types)))
if eval_types == ['proposal_fast']:
result_file = os.path.join("./result", "result_out.pkl")
recall_result, coco_recall_result, coco_precision_result, params = coco_eval(result_file, eval_types, coco)
else:
if not isinstance(results[0], dict):
out_file = os.path.join("./result", "result_out.pkl")
result_files = dict()
#faster_rcnn_r50_fpn_1x.py 走该分支 eval_types=['bbox']
if isinstance(results[0], list):
json_results = det2json(coco, results)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
mmcv.dump(json_results, result_files['bbox'])
# mask_rcnn_r50_fpn_1x.py 走该分支 eval_types=['bbox','segm']
elif isinstance(results[0], tuple):
json_results = segm2json(coco, results)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['segm'] = '{}.{}.json'.format(out_file, 'segm')
mmcv.dump(json_results[0], result_files['bbox'])
mmcv.dump(json_results[1], result_files['segm'])
elif isinstance(results[0], np.ndarray):
json_results = proposal2json(coco, results)
result_files['proposal'] = '{}.{}.json'.format(out_file, 'proposal')
mmcv.dump(json_results, result_files['proposal'])
recall_result, coco_recall_result, coco_precision_result, params = coco_eval(result_files, eval_types, coco)
else:
for name in results[0]:
out_file = os.path.join("./result", "result_out.pkl")
print('\nEvaluating {}'.format(name))
outputs_ = [out[name] for out in results]
out_file = out_file + '.{}'.format(name)
result_files = dict()
if isinstance(outputs_[0], list):
json_results = det2json(coco, outputs_)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
mmcv.dump(json_results, result_files['bbox'])
elif isinstance(outputs_[0], tuple):
json_results = segm2json(coco, outputs_)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['segm'] = '{}.{}.json'.format(out_file, 'segm')
mmcv.dump(json_results[0], result_files['bbox'])
mmcv.dump(json_results[1], result_files['segm'])
elif isinstance(outputs_[0], np.ndarray):
json_results = proposal2json(coco, outputs_)
result_files['proposal'] = '{}.{}.json'.format(out_file, 'proposal')
mmcv.dump(json_results, result_files['proposal'])
recall_result, coco_recall_result, coco_precision_result, params = coco_eval(result_files, eval_types, coco)
#只关注bbox的recall
#iouThr = None
iouThr = 0.5
print("\n", "单个类别进行评估", "\n")
coco_recall = coco_recall_result[0]
iStr = ' Categories={:>12s} | {:<18} {} @[ IoU={:<9} | area={:>6s} | maxDets={:>3d} ] = {:0.3f}'
titleStr = 'Average Recall'
typeStr = '(AR)'
iouStr = '{:0.2f}:{:0.2f}'.format(0.50, 0.95) if iouThr is None else '{:0.2f}'.format(iouThr)
areaRng = 'all'
maxDets = 100
# dimension of recall: [TxKxAxM]
s = coco_recall
if iouThr is not None:
t = np.where(iouThr == params[0].iouThrs)[0]
s = s[t]
else:
s = s[:]
for m_ind, m_cls in enumerate(categories):
temp_s = s[:, m_ind, 0, 2]
if len(temp_s[temp_s > -1]) == 0:
mean_s = -1
else:
mean_s = np.mean(temp_s[temp_s > -1])
print(iStr.format(m_cls['name'], titleStr, typeStr, iouStr, areaRng, maxDets, mean_s))
#iouThr = None
iouThr = 0.5
coco_precision = coco_precision_result[0]
iStr = ' Categories={:>12s} | {:<18} {} @[ IoU={:<9} | area={:>6s} | maxDets={:>3d} ] = {:0.3f}'
titleStr = 'Average Precision'
typeStr = '(AP)'
iouStr = '{:0.2f}:{:0.2f}'.format(0.50, 0.95) if iouThr is None else '{:0.2f}'.format(iouThr)
areaRng = 'all'
maxDets = 100
# dimension of precision: [TxRxKxAxM]
s = coco_precision
# IoU
if iouThr is not None:
t = np.where(iouThr == params[0].iouThrs)[0]
s = s[t]
else:
s = s[:]
for m_ind, m_cls in enumerate(categories):
temp_s = s[:, :, m_ind, 0, 2]
if len(temp_s[temp_s > -1]) == 0:
mean_s = -1
else:
mean_s = np.mean(temp_s[temp_s > -1])
print(iStr.format(m_cls['name'], titleStr, typeStr, iouStr, areaRng, maxDets, mean_s))
print("\n","数据分类:","\n")
#只关注bbox的recall
type_recall = np.array(recall_result[0])
catId_num, imgId_num = type_recall.shape[:2]
vis_idx = 0
num = 0
for n_key in range(imgId_num):
print(num,'/',imgId_num)
num += 1
imgId_recall = type_recall[:,n_key]
#召回率
if ((imgId_recall > 0).sum())==0:
recall = -1
else:
recall = np.sum(imgId_recall[imgId_recall > 0]) /(imgId_recall > 0).sum()
#可视化流程
result = results[n_key]
# 判断bbox和segm
if isinstance(result, tuple):
bbox_result, segm_result = result
else:
bbox_result, segm_result = result, None
#为bbox添加类别
for m_key, m_val in enumerate(bbox_result):
if m_val.shape[:2][0] > 0:
rows, clos = m_val.shape[:2]
m_temp = np.ones((rows, 1), dtype=np.float32)*m_key
bbox_result[m_key] = np.hstack((m_val, m_temp))
else:
bbox_result[m_key] = np.empty(shape=(0, 6), dtype=np.float32)
bboxes = np.vstack(bbox_result)
if score_thr > 0:
assert bboxes.shape[1] == 6
#这里需要修改-----------
#scores = bboxes[:, -1]
scores = bboxes[:, -2]
inds = scores > score_thr
bboxes = bboxes[inds, :]
# 用于通过左右方式显示原图和可视化图
img = mmcv.imread(vis_imgpath[n_key])
img = img.copy()
oriimg = img.copy()
# 画出mask
mask_list = []
if segm_result is not None:
segms = mmcv.concat_list(segm_result)
#这里需要修改-----------
#inds = np.where(bboxes[:, -1] > score_thr)[0]
inds = np.where(bboxes[:, -2] > score_thr)[0]
np.random.seed(42)
color_masks = np.random.randint(0, 256, (1, 3), dtype=np.uint8)
for i in inds:
i = int(i)
mask = maskUtils.decode(segms[i]).astype(np.bool)
img[mask] = img[mask] * 0.5 + color_masks * 0.5
mask_list.append(mask)
#画出bbox
font_scale = 0.8
thickness = 4
#bbox_color = (0, 255, 0)
#text_color = (0, 255, 0)
for bbox in bboxes:
#通过bbox[-1]获取颜色color_id = (id+1)
bbox_color = self_color[int(bbox[-1]+1)][::-1]
text_color = self_color[int(bbox[-1]+1)][::-1]
bbox_int = bbox.astype(np.int32)
left_top = (bbox_int[0], bbox_int[1])
right_bottom = (bbox_int[2], bbox_int[3])
cv2.rectangle(
img, left_top, right_bottom, bbox_color, thickness=thickness)
if len(bbox) > 4:
label_text = '{:.02f}'.format(bbox[-2])
cv2.putText(img, label_text, (bbox_int[0], bbox_int[1] - 5),
cv2.FONT_HERSHEY_COMPLEX, font_scale, text_color)
# 显示GT
annIds = coco.getAnnIds(imgIds=vis_imgid[n_key], catIds=[], iscrowd=None)
anns = coco.loadAnns(annIds)
polygons = []
color = []
category_id_list = []
for ann in anns:
if 'segmentation' in ann:
if type(ann['segmentation']) == list:
# polygon
for seg in ann['segmentation']:
poly = np.array(seg).reshape((int(len(seg) / 2), 2))
poly_list = poly.tolist()
polygons.append(poly_list)
# rgb-bgr
# mylist[start:end:step]
# 切片逆序[::-1]
if ann['iscrowd'] == 0 and ann["ignore"] == 0:
temp = self_color[ann['category_id']]
color.append(temp[::-1])
if ann['iscrowd'] == 1 or ann["ignore"] == 1:
temp = self_color[-1]
color.append(temp[::-1])
category_id_list.append(ann['category_id'])
else:
print("error type(ann['segmentation']) != list")
exit()
point_size = 2
thickness = 4
for key in range(len(polygons)):
ndata = polygons[key]
cur_color = color[key]
label_id = category_id_list[key]
label = 'error'
for m_id in categories:
if m_id['id']==label_id:
label = m_id['name']
#segmentation
if len(ndata)>2:
for k in range(len(ndata)):
data = ndata[k]
cv2.circle(oriimg, (data[0], data[1]), point_size, (cur_color[0], cur_color[1], cur_color[2]),
thickness)
# cv2.putText(oriimg, label, (int(ndata[0][0]), int(ndata[0][1])),
# cv2.FONT_HERSHEY_SIMPLEX, 0.8, (cur_color[0], cur_color[1], cur_color[2]), 2)
else:#bbox
cv2.rectangle(oriimg, (int(ndata[0][0]), int(ndata[0][1])), (int(ndata[1][0]), int(ndata[1][1])),
(cur_color[0], cur_color[1], cur_color[2]),
thickness)
cv2.putText(oriimg, label, (int(ndata[0][0]), int(ndata[0][1])),
cv2.FONT_HERSHEY_SIMPLEX, 0.8, (cur_color[0], cur_color[1], cur_color[2]), 2)
# 可视化显示mask + bbox
h1, w1 = oriimg.shape[:2]
h2, w2 = img.shape[:2]
vis = np.zeros((max(h1, h2), w1 + w2, 3), np.uint8)
vis[:h1, :w1, :] = oriimg
vis[:h2, w1:w1 + w2, :] = img
if (0< recall < 1):
print(vis_file_name[n_key])
print("------recall------:",recall," name:", 'result_{}.jpg'.format(vis_idx))
out_file = os.path.join(args.output_vis_result, "badcase", 'result_{}.jpg'.format(vis_idx))
elif (recall == -1):
out_file = os.path.join(args.output_vis_result, "abnormalcase", 'result_{}.jpg'.format(vis_idx))
else:
out_file = os.path.join(args.output_vis_result, "case", 'result_{}.jpg'.format(vis_idx))
vis_idx += 1
cv2.imwrite(out_file, vis, [int(cv2.IMWRITE_JPEG_QUALITY), 20])
if __name__ == "__main__":
'''
/mmdet/core/evaluation/recall.py
/mmdet/core/evaluation/mean_ap.py
/mmdet/core/evaluation/eval_hooks.py
/mmdet/core/evaluation/coco_utils.py
/mmdet/core/evaluation/class_names.py
/mmdet/core/evaluation/bbox_overlaps.py
/lib/python3.6/site-packages/pycocotools/cocoeval.py
/lib/python3.6/site-packages/pycocotools/coco.py
/lib/python3.6/site-packages/pycocotools/mask.py
'''
parser = argparse.ArgumentParser(
description=
"计算 mmdetection中的 mAP和AP")
parser.add_argument('-icf',
"--input_config_file",
default='/home/***/break_pad_mask_rcnn_r50_fpn_1x.py',
help="set input_config_file")
parser.add_argument('-icp',
"--input_checkpoint_file",
default='/home/***/model_break_pad_v4.pth',
help="set input_checkpoint_file")
parser.add_argument('-itj',
"--input_test_json",
default='/home/***/annotations/instances_val2017.json',
help="set input_test_json")
parser.add_argument('-itp',
"--input_test_img_path",
default='/home/***/val2017',
help="set input_test_img_path")
parser.add_argument('-ovt',
"--output_vis_result",
default='./result',
help="set output vis")
parser.add_argument('--eval',
type=str,
nargs='+',
choices=['proposal', 'proposal_fast', 'bbox', 'segm', 'keypoints'],
default=['bbox'],
help='eval types')
args = parser.parse_args()
if args.output_vis_result is None:
parser.print_help()
exit()
main(args)
temp_coco_utils.py
import mmcv
import numpy as np
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
from temp_recall import eval_recalls
def coco_eval(result_files, result_types, coco, max_dets=(100, 300, 1000)):
for res_type in result_types:
assert res_type in [
'proposal', 'proposal_fast', 'bbox', 'segm', 'keypoints'
]
if mmcv.is_str(coco):
coco = COCO(coco)
assert isinstance(coco, COCO)
if result_types == ['proposal_fast']:
ar = fast_eval_recall(result_files, coco, np.array(max_dets))
for i, num in enumerate(max_dets):
print('AR@{}\t= {:.4f}'.format(num, ar[i]))
return
recall_list = [[] for _ in range(len(result_types))]
coco_recall_list = [[] for _ in range(len(result_types))]
coco_precision_list = [[] for _ in range(len(result_types))]
p_list = []
for m_key, res_type in enumerate(result_types):
result_file = result_files[res_type]
assert result_file.endswith('.json')
coco_dets = coco.loadRes(result_file)
img_ids = coco.getImgIds()
iou_type = 'bbox' if res_type == 'proposal' else res_type
cocoEval = COCOeval(coco, coco_dets, iou_type)
cocoEval.params.imgIds = img_ids
if res_type == 'proposal':
cocoEval.params.useCats = 0
cocoEval.params.maxDets = list(max_dets)
#对给定的图像运行逐图像计算并将结果(dict列表)存储在self.evalImgs
cocoEval.evaluate()
#累积每个图像的评估结果并将结果存储到self.eval
cocoEval.accumulate()
#计算并显示评估结果,仅应用于默认参数设置
cocoEval.summarize()
#https://www.aiuai.cn/aifarm854.html
#cocoEval.eval, cocoEval.evalImgs
'''
dimension of precision: [TxRxKxAxM]
参考:https://zhuanlan.zhihu.com/p/60707912
cocoEval.eval['precision']是一个5维的数组
precision - [TxRxKxAxM] precision for every evaluation setting
catIds - [all] K cat ids to use for evaluation
iouThrs - [.5:.05:.95] T=10 IoU thresholds for evaluation
recThrs - [0:.01:1] R=101 recall thresholds for evaluation
areaRng - [...] A=4 object area ranges for evaluation
maxDets - [1 10 100] M=3 thresholds on max detections per image
第一维T:IoU的10个阈值,从0.5到0.95间隔0.05
第二维R:101个recall 阈值,从0到101
第三维K:类别,如果是想展示第一类的结果就设为0
第四维A:area 目标的大小范围 (all,small, medium, large)(全部,小,中,大)
第五维M:maxDets 单张图像中最多检测框的数量 三种 1,10,100
coco_eval.eval['precision'][0, :, 0, 0, 2] 所表示的就是当IoU=0.5时
从0到100的101个recall对应的101个precision的值
'''
#evalImgs每张图片的检测质量
#eval整个数据集上的聚合检测质量
'''
return {
'image_id': imgId,
'category_id': catId,
'aRng': aRng,
'maxDet': maxDet,
'dtIds': [d['id'] for d in dt],
'gtIds': [g['id'] for g in gt],
'dtMatches': dtm,
'gtMatches': gtm,
'dtScores': [d['score'] for d in dt],
'gtIgnore': gtIg,
'dtIgnore': dtIg,
}
self.evalImgs = [evaluateImg(imgId, catId, areaRng, maxDet)
for catId in catIds
for areaRng in p.areaRng
for imgId in p.imgIds
]
self.eval = {
'params': p,
'counts': [T, R, K, A, M],
'date': datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'),
'precision': precision,
'recall': recall,
'scores': scores,
}
'''
p = cocoEval.params
catIds = p.catIds if p.useCats else [-1]
recalls = [[-1 for _ in range(len(p.imgIds))] for _ in range(len(catIds))]
#类别
for m_catId, catId in enumerate(catIds):
#图片
for n_imgId, imgId in enumerate(p.imgIds):
if p.useCats:
gt = cocoEval._gts[imgId, catId]
dt = cocoEval._dts[imgId, catId]
else:
gt = [_ for cId in p.catIds for _ in cocoEval._gts[imgId, cId]]
dt = [_ for cId in p.catIds for _ in cocoEval._dts[imgId, cId]]
if len(gt) == 0 and len(dt) == 0:
continue
for g in gt:
if g['ignore']:
g['_ignore'] = 1
else:
g['_ignore'] = 0
# sort dt highest score first, sort gt ignore last
gtind = np.argsort([g['_ignore'] for g in gt], kind='mergesort')
gt = [gt[i] for i in gtind]
dtind = np.argsort([-d['score'] for d in dt], kind='mergesort')
#取最大的框数目100
dt = [dt[i] for i in dtind[0:100]]
iscrowd = [int(o['iscrowd']) for o in gt]
# load computed ious
ious = cocoEval.ious[imgId, catId][:, gtind] if len(cocoEval.ious[imgId, catId]) > 0 else cocoEval.ious[imgId, catId]
T = 1 # 我们只采用IOU阈值为0.5
G = len(gt)
D = len(dt)
gtm = np.zeros((T, G))
dtm = np.zeros((T, D))
gtIg = np.array([g['_ignore'] for g in gt])
dtIg = np.zeros((T, D))
if not len(ious) == 0:
recall_TPFN = len(gt)
recall_TP = 0
for dind, d in enumerate(dt):
#iou_thr = 0.5
iou_thr = 0.1
m = -1
for gind, g in enumerate(gt):
# if this gt already matched, and not a crowd, continue
# 我们只采用IOU阈值为0.5
if gtm[0,gind]>0 and not iscrowd[gind]:
continue
# if dt matched to reg gt, and on ignore gt, stop
if m>-1 and gtIg[m]==0 and gtIg[gind]==1:
break
# continue to next gt unless better match made
if ious[dind,gind] < iou_thr:
continue
# if match successful and best so far, store appropriately
iou_thr=ious[dind,gind]
m = gind
# if match made store id of match for both dt and gt
if m == -1:
continue
else:
recall_TP += 1
recalls[m_catId][n_imgId] = float(recall_TP/recall_TPFN)
#print(recalls[m_catId][n_imgId])
recall_list[m_key] = recalls
# dimension of recall: [TxKxAxM]
s1 = cocoEval.eval['recall']
# dimension of precision: [TxRxKxAxM]
s2 = cocoEval.eval['precision']
coco_recall_list[m_key] = s1
coco_precision_list[m_key] = s2
p_list.append(p)
return recall_list, coco_recall_list, coco_precision_list, p_list
def fast_eval_recall(results,
coco,
max_dets,
iou_thrs=np.arange(0.5, 0.96, 0.05)):
if mmcv.is_str(results):
assert results.endswith('.pkl')
results = mmcv.load(results)
elif not isinstance(results, list):
raise TypeError(
'results must be a list of numpy arrays or a filename, not {}'.
format(type(results)))
gt_bboxes = []
img_ids = coco.getImgIds()
for i in range(len(img_ids)):
ann_ids = coco.getAnnIds(imgIds=img_ids[i])
ann_info = coco.loadAnns(ann_ids)
if len(ann_info) == 0:
gt_bboxes.append(np.zeros((0, 4)))
continue
bboxes = []
for ann in ann_info:
if ann.get('ignore', False) or ann['iscrowd']:
continue
x1, y1, w, h = ann['bbox']
bboxes.append([x1, y1, x1 + w - 1, y1 + h - 1])
bboxes = np.array(bboxes, dtype=np.float32)
if bboxes.shape[0] == 0:
bboxes = np.zeros((0, 4))
gt_bboxes.append(bboxes)
recalls = eval_recalls(
gt_bboxes, results, max_dets, iou_thrs, print_summary=False)
ar = recalls.mean(axis=1)
return ar
def xyxy2xywh(bbox):
_bbox = bbox.tolist()
return [
_bbox[0],
_bbox[1],
_bbox[2] - _bbox[0] + 1,
_bbox[3] - _bbox[1] + 1,
]
def proposal2json(dataset, results):
json_results = []
for idx in range(len(dataset)):
img_id = dataset.img_ids[idx]
bboxes = results[idx]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = 1
json_results.append(data)
return json_results
def det2json(dataset, results):
json_results = []
for idx in range(len(dataset)):
img_id = dataset.img_ids[idx]
result = results[idx]
for label in range(len(result)):
bboxes = result[label]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = dataset.cat_ids[label]
json_results.append(data)
return json_results
def segm2json(dataset, results):
bbox_json_results = []
segm_json_results = []
for idx in range(len(dataset)):
img_id = dataset.img_ids[idx]
det, seg = results[idx]
for label in range(len(det)):
# bbox results
bboxes = det[label]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['bbox'] = xyxy2xywh(bboxes[i])
data['score'] = float(bboxes[i][4])
data['category_id'] = dataset.cat_ids[label]
bbox_json_results.append(data)
# segm results
# some detectors use different score for det and segm
if len(seg) == 2:
segms = seg[0][label]
mask_score = seg[1][label]
else:
segms = seg[label]
mask_score = [bbox[4] for bbox in bboxes]
for i in range(bboxes.shape[0]):
data = dict()
data['image_id'] = img_id
data['score'] = float(mask_score[i])
data['category_id'] = dataset.cat_ids[label]
segms[i]['counts'] = segms[i]['counts'].decode()
data['segmentation'] = segms[i]
segm_json_results.append(data)
return bbox_json_results, segm_json_results
def results2json(dataset, results, out_file):
result_files = dict()
if isinstance(results[0], list):
json_results = det2json(dataset, results)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
mmcv.dump(json_results, result_files['bbox'])
elif isinstance(results[0], tuple):
json_results = segm2json(dataset, results)
result_files['bbox'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['proposal'] = '{}.{}.json'.format(out_file, 'bbox')
result_files['segm'] = '{}.{}.json'.format(out_file, 'segm')
mmcv.dump(json_results[0], result_files['bbox'])
mmcv.dump(json_results[1], result_files['segm'])
elif isinstance(results[0], np.ndarray):
json_results = proposal2json(dataset, results)
result_files['proposal'] = '{}.{}.json'.format(out_file, 'proposal')
mmcv.dump(json_results, result_files['proposal'])
else:
raise TypeError('invalid type of results')
return result_files
temp_bbox_overlaps.py
import numpy as np
def bbox_overlaps(bboxes1, bboxes2, mode='iou'):
"""Calculate the ious between each bbox of bboxes1 and bboxes2.
Args:
bboxes1(ndarray): shape (n, 4)
bboxes2(ndarray): shape (k, 4)
mode(str): iou (intersection over union) or iof (intersection
over foreground)
Returns:
ious(ndarray): shape (n, k)
"""
assert mode in ['iou', 'iof']
bboxes1 = bboxes1.astype(np.float32)
bboxes2 = bboxes2.astype(np.float32)
rows = bboxes1.shape[0]
cols = bboxes2.shape[0]
ious = np.zeros((rows, cols), dtype=np.float32)
if rows * cols == 0:
return ious
exchange = False
if bboxes1.shape[0] > bboxes2.shape[0]:
bboxes1, bboxes2 = bboxes2, bboxes1
ious = np.zeros((cols, rows), dtype=np.float32)
exchange = True
area1 = (bboxes1[:, 2] - bboxes1[:, 0] + 1) * (
bboxes1[:, 3] - bboxes1[:, 1] + 1)
area2 = (bboxes2[:, 2] - bboxes2[:, 0] + 1) * (
bboxes2[:, 3] - bboxes2[:, 1] + 1)
for i in range(bboxes1.shape[0]):
x_start = np.maximum(bboxes1[i, 0], bboxes2[:, 0])
y_start = np.maximum(bboxes1[i, 1], bboxes2[:, 1])
x_end = np.minimum(bboxes1[i, 2], bboxes2[:, 2])
y_end = np.minimum(bboxes1[i, 3], bboxes2[:, 3])
overlap = np.maximum(x_end - x_start + 1, 0) * np.maximum(
y_end - y_start + 1, 0)
if mode == 'iou':
union = area1[i] + area2 - overlap
else:
union = area1[i] if not exchange else area2
ious[i, :] = overlap / union
if exchange:
ious = ious.T
return ious
temp_recall.py
import numpy as np
from terminaltables import AsciiTable
from temp_bbox_overlaps import bbox_overlaps
def _recalls(all_ious, proposal_nums, thrs):
img_num = all_ious.shape[0]
total_gt_num = sum([ious.shape[0] for ious in all_ious])
_ious = np.zeros((proposal_nums.size, total_gt_num), dtype=np.float32)
for k, proposal_num in enumerate(proposal_nums):
tmp_ious = np.zeros(0)
for i in range(img_num):
ious = all_ious[i][:, :proposal_num].copy()
gt_ious = np.zeros((ious.shape[0]))
if ious.size == 0:
tmp_ious = np.hstack((tmp_ious, gt_ious))
continue
for j in range(ious.shape[0]):
gt_max_overlaps = ious.argmax(axis=1)
max_ious = ious[np.arange(0, ious.shape[0]), gt_max_overlaps]
gt_idx = max_ious.argmax()
gt_ious[j] = max_ious[gt_idx]
box_idx = gt_max_overlaps[gt_idx]
ious[gt_idx, :] = -1
ious[:, box_idx] = -1
tmp_ious = np.hstack((tmp_ious, gt_ious))
_ious[k, :] = tmp_ious
_ious = np.fliplr(np.sort(_ious, axis=1))
recalls = np.zeros((proposal_nums.size, thrs.size))
for i, thr in enumerate(thrs):
recalls[:, i] = (_ious >= thr).sum(axis=1) / float(total_gt_num)
return recalls
def set_recall_param(proposal_nums, iou_thrs):
"""Check proposal_nums and iou_thrs and set correct format.
"""
if isinstance(proposal_nums, list):
_proposal_nums = np.array(proposal_nums)
elif isinstance(proposal_nums, int):
_proposal_nums = np.array([proposal_nums])
else:
_proposal_nums = proposal_nums
if iou_thrs is None:
_iou_thrs = np.array([0.5])
elif isinstance(iou_thrs, list):
_iou_thrs = np.array(iou_thrs)
elif isinstance(iou_thrs, float):
_iou_thrs = np.array([iou_thrs])
else:
_iou_thrs = iou_thrs
return _proposal_nums, _iou_thrs
def eval_recalls(gts,
proposals,
proposal_nums=None,
iou_thrs=None,
print_summary=True):
"""Calculate recalls.
Args:
gts(list or ndarray): a list of arrays of shape (n, 4)
proposals(list or ndarray): a list of arrays of shape (k, 4) or (k, 5)
proposal_nums(int or list of int or ndarray): top N proposals
thrs(float or list or ndarray): iou thresholds
Returns:
ndarray: recalls of different ious and proposal nums
"""
img_num = len(gts)
assert img_num == len(proposals)
proposal_nums, iou_thrs = set_recall_param(proposal_nums, iou_thrs)
all_ious = []
for i in range(img_num):
if proposals[i].ndim == 2 and proposals[i].shape[1] == 5:
scores = proposals[i][:, 4]
sort_idx = np.argsort(scores)[::-1]
img_proposal = proposals[i][sort_idx, :]
else:
img_proposal = proposals[i]
prop_num = min(img_proposal.shape[0], proposal_nums[-1])
if gts[i] is None or gts[i].shape[0] == 0:
ious = np.zeros((0, img_proposal.shape[0]), dtype=np.float32)
else:
ious = bbox_overlaps(gts[i], img_proposal[:prop_num, :4])
all_ious.append(ious)
all_ious = np.array(all_ious)
recalls = _recalls(all_ious, proposal_nums, iou_thrs)
if print_summary:
print_recall_summary(recalls, proposal_nums, iou_thrs)
return recalls
def print_recall_summary(recalls,
proposal_nums,
iou_thrs,
row_idxs=None,
col_idxs=None):
"""Print recalls in a table.
Args:
recalls(ndarray): calculated from `bbox_recalls`
proposal_nums(ndarray or list): top N proposals
iou_thrs(ndarray or list): iou thresholds
row_idxs(ndarray): which rows(proposal nums) to print
col_idxs(ndarray): which cols(iou thresholds) to print
"""
proposal_nums = np.array(proposal_nums, dtype=np.int32)
iou_thrs = np.array(iou_thrs)
if row_idxs is None:
row_idxs = np.arange(proposal_nums.size)
if col_idxs is None:
col_idxs = np.arange(iou_thrs.size)
row_header = [''] + iou_thrs[col_idxs].tolist()
table_data = [row_header]
for i, num in enumerate(proposal_nums[row_idxs]):
row = [
'{:.3f}'.format(val)
for val in recalls[row_idxs[i], col_idxs].tolist()
]
row.insert(0, num)
table_data.append(row)
table = AsciiTable(table_data)
print(table.table)
def plot_num_recall(recalls, proposal_nums):
"""Plot Proposal_num-Recalls curve.
Args:
recalls(ndarray or list): shape (k,)
proposal_nums(ndarray or list): same shape as `recalls`
"""
if isinstance(proposal_nums, np.ndarray):
_proposal_nums = proposal_nums.tolist()
else:
_proposal_nums = proposal_nums
if isinstance(recalls, np.ndarray):
_recalls = recalls.tolist()
else:
_recalls = recalls
import matplotlib.pyplot as plt
f = plt.figure()
plt.plot([0] + _proposal_nums, [0] + _recalls)
plt.xlabel('Proposal num')
plt.ylabel('Recall')
plt.axis([0, proposal_nums.max(), 0, 1])
f.show()
def plot_iou_recall(recalls, iou_thrs):
"""Plot IoU-Recalls curve.
Args:
recalls(ndarray or list): shape (k,)
iou_thrs(ndarray or list): same shape as `recalls`
"""
if isinstance(iou_thrs, np.ndarray):
_iou_thrs = iou_thrs.tolist()
else:
_iou_thrs = iou_thrs
if isinstance(recalls, np.ndarray):
_recalls = recalls.tolist()
else:
_recalls = recalls
import matplotlib.pyplot as plt
f = plt.figure()
plt.plot(_iou_thrs + [1.0], _recalls + [0.])
plt.xlabel('IoU')
plt.ylabel('Recall')
plt.axis([iou_thrs.min(), 1, 0, 1])
f.show()