chore(local-tool): 移除模型 FPS/latency/accuracy 預估值

使用者要求拿掉 FPS 預估數字(未經實測,不準確)。

- docs/LOCAL-TOOL-SPEC.md: 模型表格移除 FPS 欄位
- server/data/models.json: 7 個模型全部移除 fps / latencyMs / accuracy 欄位

前端 model-detail / model-card 有讀這些欄位的 UI,移除後會顯示 — 或不顯示
該列,不需要額外前端改動(已有 null/undefined fallback)。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
jim800121chen 2026-04-16 15:01:25 +08:00
parent 61f9b8bf6b
commit 4f9a193782
2 changed files with 163 additions and 60 deletions

View File

@ -126,20 +126,20 @@ visionA Local Tool 是 Kneron KL520 / KL720 邊緣 AI 推論硬體的**本機桌
### KL5204 個) ### KL5204 個)
| 模型 ID | 名稱 | 任務 | 輸入 | FPS | | 模型 ID | 名稱 | 任務 | 輸入 |
|---------|------|------|------|-----| |---------|------|------|------|
| kl520-yolov5-detection | YOLOv5 Detection | 物件偵測 | 640×640 | ~20 | | kl520-yolov5-detection | YOLOv5 Detection | 物件偵測 | 640×640 |
| kl520-fcos-detection | FCOS Detection | 物件偵測 | 512×512 | ~22 | | kl520-fcos-detection | FCOS Detection | 物件偵測 | 512×512 |
| kl520-ssd-face-detection | SSD Face Detection | 人臉偵測 | 320×240 | ~100 | | kl520-ssd-face-detection | SSD Face Detection | 人臉偵測 | 320×240 |
| kl520-tiny-yolov3 | Tiny YOLOv3 | 物件偵測 | 416×416 | ~28 | | kl520-tiny-yolov3 | Tiny YOLOv3 | 物件偵測 | 416×416 |
### KL7203 個) ### KL7203 個)
| 模型 ID | 名稱 | 任務 | 輸入 | FPS | | 模型 ID | 名稱 | 任務 | 輸入 |
|---------|------|------|------|-----| |---------|------|------|------|
| kl720-yolov5-detection | YOLOv5 Detection | 物件偵測 | 640×640 | ~33 | | kl720-yolov5-detection | YOLOv5 Detection | 物件偵測 | 640×640 |
| kl720-resnet18-classification | ResNet18 Classification | 分類1000 類)| 224×224 | ~100 | | kl720-resnet18-classification | ResNet18 Classification | 分類1000 類)| 224×224 |
| kl720-fcos-detection | FCOS Detection | 物件偵測 | 512×512 | ~33 | | kl720-fcos-detection | FCOS Detection | 物件偵測 | 512×512 |
--- ---

View File

@ -5,16 +5,32 @@
"description": "YOLOv5 object detection model compiled for Kneron KL520. No upsample variant optimized for NPU inference at 640x640 resolution.", "description": "YOLOv5 object detection model compiled for Kneron KL520. No upsample variant optimized for NPU inference at 640x640 resolution.",
"thumbnail": "/images/models/yolov5-det.png", "thumbnail": "/images/models/yolov5-det.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["general", "multi-object"], "categories": [
"general",
"multi-object"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 640, "height": 640}, "inputSize": {
"width": 640,
"height": 640
},
"modelSize": 7200000, "modelSize": 7200000,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.80, "supportedHardware": [
"latencyMs": 50, "KL520"
"fps": 20, ],
"supportedHardware": ["KL520"], "labels": [
"labels": ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light"], "person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light"
],
"filePath": "data/nef/kl520/kl520_20005_yolov5-noupsample_w640h640.nef", "filePath": "data/nef/kl520/kl520_20005_yolov5-noupsample_w640h640.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -28,16 +44,32 @@
"description": "FCOS (Fully Convolutional One-Stage) object detection with DarkNet53s backbone, compiled for KL520. Anchor-free detection at 512x512.", "description": "FCOS (Fully Convolutional One-Stage) object detection with DarkNet53s backbone, compiled for KL520. Anchor-free detection at 512x512.",
"thumbnail": "/images/models/fcos-det.png", "thumbnail": "/images/models/fcos-det.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["general", "multi-object"], "categories": [
"general",
"multi-object"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 512, "height": 512}, "inputSize": {
"width": 512,
"height": 512
},
"modelSize": 8900000, "modelSize": 8900000,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.78, "supportedHardware": [
"latencyMs": 45, "KL520"
"fps": 22, ],
"supportedHardware": ["KL520"], "labels": [
"labels": ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light"], "person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light"
],
"filePath": "data/nef/kl520/kl520_20004_fcos-drk53s_w512h512.nef", "filePath": "data/nef/kl520/kl520_20004_fcos-drk53s_w512h512.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -51,16 +83,23 @@
"description": "SSD-based face detection with landmark localization, compiled for KL520. Lightweight model suitable for face detection and alignment tasks.", "description": "SSD-based face detection with landmark localization, compiled for KL520. Lightweight model suitable for face detection and alignment tasks.",
"thumbnail": "/images/models/ssd-face.png", "thumbnail": "/images/models/ssd-face.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["face", "security"], "categories": [
"face",
"security"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 320, "height": 240}, "inputSize": {
"width": 320,
"height": 240
},
"modelSize": 1000000, "modelSize": 1000000,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.85, "supportedHardware": [
"latencyMs": 10, "KL520"
"fps": 100, ],
"supportedHardware": ["KL520"], "labels": [
"labels": ["face"], "face"
],
"filePath": "data/nef/kl520/kl520_ssd_fd_lm.nef", "filePath": "data/nef/kl520/kl520_ssd_fd_lm.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -74,16 +113,32 @@
"description": "Tiny YOLOv3 object detection model compiled for KL520. Compact and fast model for general-purpose multi-object detection on edge devices.", "description": "Tiny YOLOv3 object detection model compiled for KL520. Compact and fast model for general-purpose multi-object detection on edge devices.",
"thumbnail": "/images/models/tiny-yolov3.png", "thumbnail": "/images/models/tiny-yolov3.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["general", "multi-object"], "categories": [
"general",
"multi-object"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 416, "height": 416}, "inputSize": {
"width": 416,
"height": 416
},
"modelSize": 9400000, "modelSize": 9400000,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.75, "supportedHardware": [
"latencyMs": 35, "KL520"
"fps": 28, ],
"supportedHardware": ["KL520"], "labels": [
"labels": ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light"], "person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light"
],
"filePath": "data/nef/kl520/kl520_tiny_yolo_v3.nef", "filePath": "data/nef/kl520/kl520_tiny_yolo_v3.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -97,16 +152,32 @@
"description": "YOLOv5 object detection model compiled for Kneron KL720. No upsample variant optimized for KL720 NPU inference at 640x640 resolution with USB 3.0 throughput.", "description": "YOLOv5 object detection model compiled for Kneron KL720. No upsample variant optimized for KL720 NPU inference at 640x640 resolution with USB 3.0 throughput.",
"thumbnail": "/images/models/yolov5-det.png", "thumbnail": "/images/models/yolov5-det.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["general", "multi-object"], "categories": [
"general",
"multi-object"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 640, "height": 640}, "inputSize": {
"width": 640,
"height": 640
},
"modelSize": 10168348, "modelSize": 10168348,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.82, "supportedHardware": [
"latencyMs": 30, "KL720"
"fps": 33, ],
"supportedHardware": ["KL720"], "labels": [
"labels": ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light"], "person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light"
],
"filePath": "data/nef/kl720/kl720_20005_yolov5-noupsample_w640h640.nef", "filePath": "data/nef/kl720/kl720_20005_yolov5-noupsample_w640h640.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -120,16 +191,32 @@
"description": "ResNet18-based image classification compiled for KL720. Supports 1000 ImageNet categories with fast inference via USB 3.0.", "description": "ResNet18-based image classification compiled for KL720. Supports 1000 ImageNet categories with fast inference via USB 3.0.",
"thumbnail": "/images/models/imagenet-cls.png", "thumbnail": "/images/models/imagenet-cls.png",
"taskType": "classification", "taskType": "classification",
"categories": ["general", "image-classification"], "categories": [
"general",
"image-classification"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 224, "height": 224}, "inputSize": {
"width": 224,
"height": 224
},
"modelSize": 12826804, "modelSize": 12826804,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.78, "supportedHardware": [
"latencyMs": 10, "KL720"
"fps": 100, ],
"supportedHardware": ["KL720"], "labels": [
"labels": ["airplane", "automobile", "bird", "cat", "deer", "dog", "frog", "horse", "ship", "truck"], "airplane",
"automobile",
"bird",
"cat",
"deer",
"dog",
"frog",
"horse",
"ship",
"truck"
],
"filePath": "data/nef/kl720/kl720_20001_resnet18_w224h224.nef", "filePath": "data/nef/kl720/kl720_20001_resnet18_w224h224.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",
@ -143,16 +230,32 @@
"description": "FCOS (Fully Convolutional One-Stage) object detection with DarkNet53s backbone, compiled for KL720. Anchor-free detection at 512x512.", "description": "FCOS (Fully Convolutional One-Stage) object detection with DarkNet53s backbone, compiled for KL720. Anchor-free detection at 512x512.",
"thumbnail": "/images/models/fcos-det.png", "thumbnail": "/images/models/fcos-det.png",
"taskType": "object_detection", "taskType": "object_detection",
"categories": ["general", "multi-object"], "categories": [
"general",
"multi-object"
],
"framework": "NEF", "framework": "NEF",
"inputSize": {"width": 512, "height": 512}, "inputSize": {
"width": 512,
"height": 512
},
"modelSize": 13004640, "modelSize": 13004640,
"quantization": "INT8", "quantization": "INT8",
"accuracy": 0.80, "supportedHardware": [
"latencyMs": 30, "KL720"
"fps": 33, ],
"supportedHardware": ["KL720"], "labels": [
"labels": ["person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light"], "person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light"
],
"filePath": "data/nef/kl720/kl720_20004_fcos-drk53s_w512h512.nef", "filePath": "data/nef/kl720/kl720_20004_fcos-drk53s_w512h512.nef",
"version": "1.0.0", "version": "1.0.0",
"author": "Kneron", "author": "Kneron",