https://github.com/shuai-xie/asm_web
ASM web inferface
https://github.com/shuai-xie/asm_web
active-learning detection flask
Last synced: 3 months ago
JSON representation
ASM web inferface
- Host: GitHub
- URL: https://github.com/shuai-xie/asm_web
- Owner: Shuai-Xie
- Created: 2020-02-19T07:25:18.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2023-05-01T21:20:28.000Z (about 3 years ago)
- Last Synced: 2025-01-11T04:44:09.165Z (over 1 year ago)
- Topics: active-learning, detection, flask
- Language: JavaScript
- Homepage:
- Size: 819 KB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# ASM_web
Flask Web demo with [ASM_core](https://github.com/Shuai-Xie/ASM_core).
## Features
ASM steps with Dataturks.
**1. auto_label()**
- update `d_hits.status` by ASM detection results.
```py
# update image status in Dataturks d_hits table
status = 'sl' if len(sl_boxes) > 0 else 'al'
# store all the sl anns, to compute AL ratio in start_train()
global sl_anns = []
```
**2. Human annotation in Dataturks Hits Done, filter by SL/AL status.**
**3. start_train()**
- read `gt SL/AL anns` from mysql
- compute AL ratio by `sl_anns` and `gt_sl_anns`
- choose topK and build `asm_train_anns`
- train on the new dataset
```py
# 从数据库查询 3 种不同 status 标注
gt_sl_anns = get_gt_anns_from_sql(dataset, names2class, status='sl')
gt_al_anns = get_gt_anns_from_sql(dataset, names2class, status='al')
label_anns = get_gt_anns_from_sql(dataset, names2class, status='done')
```
## Architecture
