Outline possible Tasks

Approach with a client SDK route over Framework route.

- [ ] Generation
- [ ] Diarization
- [ ] Transcription
- [ ] Categorization
- [ ] Embeddings
- [ ] Captioning
- [ ] Object Detection
- [ ] Dense Region Captioning
- [ ] Region Proposal
- [ ] Caption-to-Phrase Grounding
- [ ] Referring Expression Segmentation
- [ ] Region-to-Segmentation
- [ ] Open Vocabulary Detection
- [ ] Region-to-Category Classification
- [ ] Region-to-Description
- [ ] Optical Character Recognition
- [ ] Chunking
- [ ] Audio Extraction
- [ ] Key Frame Extraction
- [ ] Face Extraction

該 issue 列出了許多 AI/ML 任務，如說話人日誌、轉錄和物件偵測。首先探索儲存庫結構以了解現有程式碼庫和資料流。確定目前處理類似任務的模組或服務，並檢視任何相關文件或範例。「完成」將涉及為其中一項或多項任務設計和實作客戶端 SDK 方法，但範圍極其廣泛且未定義。

由索引模型根據 Issue 內容生成。