feat: 事件时间轴缩略帧 + 人物管理头像 + 人物合并硬规则校验

## 新架构:Oracle 集中计算 + NAS 代理展示

### Oracle 端 (fam-edge)
- 新增 frame_service: ffmpeg 视频抽帧 + VLM 人物定位裁剪头像(磁盘缓存)
- 新增 /api/oracle/frame: 按 video_id+ts 抽帧返回 jpeg(带 token)
- 新增 /api/oracle/avatar: 按 label 生成人物头像(VLM 定位人物 + 兜底整帧居中)
- 新增 person_identifier: 人物身份识别模块
- Gemini 适配器支持 flash/flash-lite 双模型切换,429 自动降级
- frame_service VLM 全模型 429 时进入 10 分钟熔断,避免每次请求白打配额
- 兜底头像不落缓存,配额恢复后自动重试 VLM 精确定位

### 人物合并硬规则校验(框架级修复)
- person_service: LLM 合并结果落库前加硬冲突检测
  - 性别冲突 → 绝不合并
  - 年龄档跨未成年/成年 → 绝不合并(防止把爷爷/宝宝并进同一人)
- oracle_db: upsert_person 入口剥离括号后缀(人物A(别名:人物B) → 人物A),消灭垃圾人物行
- 修复 set_canonical 丢弃 source 参数的 bug(旧代码硬编码 'manual' 导致错误合并被永久固化)
- get_events_for_label: 只提取该身份组的特征文本,头像定位更精准

### NAS 端 (fam-core)
- 新增 img_proxy: /api/proxy/frame 和 /api/proxy/avatar 代理 Oracle 图片
- app.py 注册 img_bp 蓝图
- oracle_sync / db_layer / member_manager 同步人物表

### UI 端 (fam-ui)
- 事件时间轴: 每条事件卡片加时间点缩略帧
- 人物管理: 每人卡片加头像(150x150 圆角)
- parse_persons: 剥离括号备注,与 Oracle 归一化一致
- 新增 EventItem 组件、Timeline 页改造
- Chat / ServiceStatus 页相应调整

### 数据库
- scripts/ddl.sql: 同步表结构更新
- Oracle people 表: features_json / display_uid / source 字段完善
This commit is contained in:
ericwyuan
2026-08-23 00:13:56 +08:00
parent aca1a674b1
commit 2c5bf950c5
31 changed files with 2166 additions and 25 deletions

View File

@@ -1,12 +1,39 @@
<script setup>
import { computed } from 'vue'
import { fmtTime, parsePersons } from '../api.js'
import { computed, ref } from 'vue'
import { api, fmtTime, parsePersons } from '../api.js'
import Badge from './Badge.vue'
const props = defineProps({
event: { type: Object, required: true },
videoId: { type: [Number, String], required: true },
})
const emit = defineEmits(['corrected'])
// 闭集人物识别只有这 4 个真实成员,纠错时二选/四选一,不是自由填字符串
const CANONICAL_NAMES = ['爷爷', '爸爸', '媳妇', '汤圆']
const fixingPerson = ref(null) // 当前正在纠错的人物名(打开选择器)
const fixingBusy = ref(false)
const fixError = ref('')
function toggleFix(name) {
fixError.value = ''
fixingPerson.value = fixingPerson.value === name ? null : name
}
async function correctPerson(oldName, newName) {
if (oldName === newName) { fixingPerson.value = null; return }
fixingBusy.value = true
fixError.value = ''
try {
await api.identityCorrect(props.videoId, oldName, newName)
fixingPerson.value = null
emit('corrected')
} catch (e) {
fixError.value = e.message
} finally {
fixingBusy.value = false
}
}
const timeLabel = computed(() => fmtTime(props.event.ts))
const persons = computed(() => parsePersons(props.event.person_list_json))
@@ -49,10 +76,27 @@ const appearances = computed(() => {
<div v-if="thumbUrl" class="mb-2.5 leading-none">
<img loading="lazy" alt="事件帧" :src="thumbUrl" class="block w-full rounded-lg border border-border" />
</div>
<div class="mb-2 flex flex-wrap gap-2">
<div class="mb-2 flex flex-wrap items-start gap-2">
<Badge v-if="isAttention" tone="danger"> 需关注</Badge>
<Badge v-for="p in persons" :key="p" tone="accent">{{ p }}</Badge>
<div v-for="p in persons" :key="p" class="relative">
<button type="button" class="group inline-flex items-center gap-1" @click="toggleFix(p)" title="点击纠正这个人是谁">
<Badge tone="accent">{{ p }}</Badge>
<span class="text-[10px] text-text-faint opacity-0 transition-opacity group-hover:opacity-100"></span>
</button>
<div v-if="fixingPerson === p"
class="absolute left-0 top-full z-10 mt-1 flex gap-1.5 rounded-lg border border-border-hi bg-panel p-2 shadow-[var(--shadow-card)]">
<button v-for="name in CANONICAL_NAMES" :key="name" :disabled="fixingBusy"
@click="correctPerson(p, name)"
class="rounded-md border px-2 py-1 text-xs disabled:opacity-50"
:class="name === p
? 'border-accent/50 bg-accent/15 text-accent-bright'
: 'border-border bg-panel-2 text-text-dim hover:border-accent/40 hover:text-white'">
{{ name }}
</button>
</div>
</div>
</div>
<div v-if="fixError" class="mb-2 text-xs text-danger">纠错失败{{ fixError }}</div>
<div v-if="appearances.length" class="mb-2">
<div v-for="a in appearances" :key="a.uid" class="my-1.5 rounded-lg border border-border bg-panel-3 px-2.5 py-1.5 text-xs">
<b class="text-accent-bright">{{ a.uid }}</b>