38 lines
1.1 KiB
Python
38 lines
1.1 KiB
Python
|
import json
|
||
|
|
import numpy as np
|
||
|
|
from collections import defaultdict
|
||
|
|
|
||
|
|
data_path = 'eval.json.bak'
|
||
|
|
with open(data_path, 'r') as f:
|
||
|
|
data = json.load(f)
|
||
|
|
|
||
|
|
|
||
|
|
video_ids = data['video_id']
|
||
|
|
frame_ids = data['frame_id']
|
||
|
|
loss_gious = data['loss_giou']
|
||
|
|
record = {}
|
||
|
|
for video_id, frame_id, loss_giou in zip(video_ids, frame_ids, loss_gious):
|
||
|
|
for v_id, f_id, loss in zip(video_id, frame_id, loss_giou):
|
||
|
|
if v_id[0] not in record:
|
||
|
|
record[v_id[0]] = {'frame_id': [], 'loss_giou': []}
|
||
|
|
if f_id[0] not in record[v_id[0]]['frame_id']:
|
||
|
|
record[v_id[0]]['frame_id'].append(f_id[0])
|
||
|
|
record[v_id[0]]['loss_giou'].append(loss)
|
||
|
|
|
||
|
|
videos = []
|
||
|
|
avg_loss = []
|
||
|
|
for v_id in record.keys():
|
||
|
|
f_ids = record[v_id]['frame_id']
|
||
|
|
loss = record[v_id]['loss_giou']
|
||
|
|
videos.append(v_id)
|
||
|
|
avg_loss.append(np.array(loss).mean())
|
||
|
|
print(f"video id: {v_id}, loss: {np.array(loss).mean()}")
|
||
|
|
|
||
|
|
avg_loss = np.array(avg_loss)
|
||
|
|
order_ids = np.argsort(avg_loss)
|
||
|
|
|
||
|
|
print("The top N best video id")
|
||
|
|
for index in order_ids[0:40]:
|
||
|
|
print(f"video id: {videos[index]}, loss: {avg_loss[index]}")
|
||
|
|
|