def finalize(self, result_dict, ds=None, output_dir=None):
result_dict["good"] += self.good
result_dict["total"] += self.total
detections = np.array(self.results)
image_ids = []
for idx in range(0, detections.shape[0]):
# this is the index into the image list
image_id = int(detections[idx][0])
image_ids.append(image_id)
# map it to the coco image it
detections[idx][0] = ds.image_ids[image_id]
height, width = ds.image_sizes[image_id]
# box comes from model as: ymin, xmin, ymax, xmax
ymin = detections[idx][1] * height
xmin = detections[idx][2] * width
ymax = detections[idx][3] * height
xmax = detections[idx][4] * width
# from pycoco wants {imageID,x1,y1,w,h,score,class}
detections[idx][1] = xmin
detections[idx][2] = ymin
detections[idx][3] = xmax - xmin
detections[idx][4] = ymax - ymin
# for debugging
if True:
pp = []
for idx in range(0, detections.shape[0]):
pp.append({
"image_id":
int(detections[idx][0]),
"image_loc":
ds.get_item_loc(image_ids[idx]),
"category_id":
int(detections[idx][6]),
"bbox": [
float(detections[idx][1]),
float(detections[idx][2]),
float(detections[idx][3]),
float(detections[idx][4])
],
"score":
float(detections[idx][5])
})
if not output_dir:
output_dir = "/tmp"
fname = "{}/{}.json".format(output_dir, result_dict["scenario"])
with open(fname, "w") as fp:
json.dump(pp, fp, sort_keys=True, indent=4)
image_ids = list(set([i[0] for i in detections]))
self.results = []
cocoGt = pycoco.COCO(ds.annotation_file)
cocoDt = cocoGt.loadRes(detections)
cocoEval = COCOeval(cocoGt, cocoDt, iouType='bbox')
cocoEval.params.imgIds = image_ids
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
result_dict["mAP"] = cocoEval.stats[0]
def finalize(self, result_dict, ds=None, output_dir=None):
result_dict["good"] += self.good
result_dict["total"] += self.total
if self.use_inv_map:
# for pytorch
label_map = {}
with open(ds.annotation_file) as fin:
annotations = json.load(fin)
for cnt, cat in enumerate(annotations["categories"]):
label_map[cat["id"]] = cnt + 1
inv_map = {v: k for k, v in label_map.items()}
detections = []
image_indices = []
for batch in range(0, len(self.results)):
image_indices.append(self.content_ids[batch])
for idx in range(0, len(self.results[batch])):
detection = self.results[batch][idx]
# this is the index of the coco image
image_idx = int(detection[0])
if image_idx != self.content_ids[batch]:
# working with the coco index/id is error prone - extra check to make sure it is consistent
log.error("image_idx missmatch, lg={} / result={}".format(
image_idx, self.content_ids[batch]))
# map the index to the coco image id
detection[0] = ds.image_ids[image_idx]
height, width = ds.image_sizes[image_idx]
# box comes from model as: ymin, xmin, ymax, xmax
ymin = detection[1] * height
xmin = detection[2] * width
ymax = detection[3] * height
xmax = detection[4] * width
# pycoco wants {imageID,x1,y1,w,h,score,class}
detection[1] = xmin
detection[2] = ymin
detection[3] = xmax - xmin
detection[4] = ymax - ymin
if self.use_inv_map:
cat_id = inv_map.get(int(detection[6]), -1)
if cat_id == -1:
# FIXME:
log.info("finalize can't map category {}".format(
int(detection[6])))
detection[6] = cat_id
detections.append(np.array(detection))
# map indices to coco image id's
image_ids = [ds.image_ids[i] for i in image_indices]
self.results = []
cocoGt = pycoco.COCO(ds.annotation_file)
cocoDt = cocoGt.loadRes(np.array(detections))
cocoEval = COCOeval(cocoGt, cocoDt, iouType='bbox')
cocoEval.params.imgIds = image_ids
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
result_dict["mAP"] = cocoEval.stats[0]
def finalize(self, result_dict, ds=None):
result_dict["good"] += self.good
result_dict["total"] += self.total
detections = np.array(self.results)
for idx in range(0, detections.shape[0]):
image_idx = int(detections[idx][0])
image_id = ds.image_ids[image_idx]
detections[idx][0] = image_id
height, width = ds.image_sizes[image_idx]
# box comes from model as: ymin, xmin, ymax, xmax
ymin = detections[idx][1] * height
xmin = detections[idx][2] * width
ymax = detections[idx][3] * height
xmax = detections[idx][4] * width
# from pycoco wants {imageID,x1,y1,w,h,score,class}
detections[idx][1] = xmin
detections[idx][2] = ymin
detections[idx][3] = xmax - xmin
detections[idx][4] = ymax - ymin
if False:
# TODO: remove after debug
pp = []
for idx in range(0, detections.shape[0]):
pp.append({
"image_id":
int(detections[idx][0]),
"category_id":
int(detections[idx][6]),
"bbox": [
float(detections[idx][1]),
float(detections[idx][2]),
float(detections[idx][3]),
float(detections[idx][4])
],
"score":
float(detections[idx][5])
})
import json
with open("/tmp/t.json", "w") as fp:
json.dump(pp, fp, sort_keys=True, indent=4)
image_ids = list(set([i[0] for i in detections]))
self.results = []
cocoGt = pycoco.COCO(ds.annotation_file)
cocoDt = cocoGt.loadRes(detections)
cocoEval = COCOeval(cocoGt, cocoDt, iouType='bbox')
cocoEval.params.imgIds = image_ids
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
result_dict["mAP"] = cocoEval.stats[0]
def finalize(self, result_dict, ds=None, output_dir=None):
result_dict["good"] += self.good
result_dict["total"] += self.total
if self.use_inv_map:
# for pytorch
label_map = {}
with open(ds.annotation_file) as fin:
annotations = json.load(fin)
for cnt, cat in enumerate(annotations["categories"]):
label_map[cat["id"]] = cnt + 1
inv_map = {v:k for k,v in label_map.items()}
detections = []
image_indices = []
for batch in range(0, len(self.results)):
for idx in range(0, len(self.results[batch])):
detection = self.results[batch][idx]
# this is the index of the coco image
image_idx = int(detection[0])
# because we need to have the
image_indices.append(image_idx)
# map the index to the coco image id
detection[0] = ds.image_ids[image_idx]
height, width = ds.image_sizes[image_idx]
# box comes from model as: ymin, xmin, ymax, xmax
ymin = detection[1] * height
xmin = detection[2] * width
ymax = detection[3] * height
xmax = detection[4] * width
# pycoco wants {imageID,x1,y1,w,h,score,class}
detection[1] = xmin
detection[2] = ymin
detection[3] = xmax - xmin
detection[4] = ymax - ymin
if self.use_inv_map:
cat_id = inv_map.get(int(detection[6]), -1)
if cat_id == -1:
# FIXME:
log.info("finalize can't map category {}".format(int(detection[6])))
detection[6] = cat_id
detections.append(np.array(detection))
if output_dir:
# for debugging
pp = []
for image_idx, detection in zip(image_indices, detections):
pp.append({"image_id": int(detection[0]),
"image_loc": ds.get_item_loc(image_idx),
"category_id": int(detection[6]),
"bbox": [float(detection[1]), float(detection[2]),
float(detection[3]), float(detection[4])],
"score": float(detection[5])})
fname = "{}.json".format(result_dict["scenario"])
with open(fname, "w") as fp:
json.dump(pp, fp, sort_keys=True, indent=4)
image_ids = list(set([i[0] for i in detections]))
self.results = []
cocoGt = pycoco.COCO(ds.annotation_file)
cocoDt = cocoGt.loadRes(np.array(detections))
cocoEval = COCOeval(cocoGt, cocoDt, iouType='bbox')
cocoEval.params.imgIds = image_ids
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
result_dict["mAP"] = cocoEval.stats[0]