Total runs:
16.5K
Run Growth:
6.5K
Growth Rate:
39.28%
Sponsored by
APIMart
mlboydaisuke / RF-DETR-CoreAI
Model's Last Updated: July 24 2026
object-detection
Introduction of RF-DETR-CoreAI
Model Details of RF-DETR-CoreAI
RF-DETR — Core AI (
.aimodel
)
RF-DETR (Roboflow's real-time detection transformer, COCO-pretrained) converted to Apple Core AI for iOS 27 / macOS 27 — the answer to apple/coreai-models#14 . DETR family = no NMS : post-processing is one sigmoid + top-k.
Files
| file | input | params | M4 Max GPU | iPhone 17 Pro GPU |
|---|---|---|---|---|
rfdetr-nano_float32.aimodel
|
384×384 | 30.5M | 8.6 ms (~116 FPS) | ~25 ms (33–39 FPS live) |
rfdetr-small_float32.aimodel
|
512×512 | 32.1M | 12.0 ms (~83 FPS) | — |
rfdetr-medium_float32.aimodel
|
576×576 | 33.7M | 14.8 ms (~68 FPS) | 56–63 ms (15–17 FPS live) |
rfdetr-large_float32.aimodel
|
704×704 | 33.9M | 19.1 ms (~52 FPS) | — |
iPhone numbers are end-to-end live-camera measurements from the CoreAIKit DetectCamera example (Release; zero-copy capture pipeline — AVCaptureVideoPreviewLayer display, hardware-scaled 32BGRA buffers, vImage preprocessing overlapped with GPU inference). Peak measured 39.6 FPS ≈ the nano model ceiling; sustained max-load throughput drops on a hot chassis (thermal).
fp32 is the ship dtype: it gates detection-set exact vs the PyTorch fp32 reference on CPU and GPU (per confident detection: same class, IoU ≥ 0.999 measured, score within 2e-3), and fp16 only bought ~7% latency on M4 Max while adding near-tie ranking noise.
Graph contract
input "image" [1, 3, R, R] float32, RGB in [0, 1] (ImageNet mean/std folded in-graph)
output "dets" [1, 300, 4] boxes, cxcywh normalized to [0, 1]
output "labels" [1, 300, 91] raw class logits; column index = ORIGINAL COCO id (0 unused, 1=person … 17=cat … 90)
Python decode sketch (Swift is the same three steps):
import numpy as np, coreai.runtime as rt
model = await rt.AIModel.load(path, rt.SpecializationOptions.default())
fn = model.load_function("main")
out = await fn({"image": rt.NDArray(rgb01)}) # rgb01: [1,3,R,R] in [0,1]
prob = 1 / (1 + np.exp(-out["labels"].numpy()[0])) # [300, 91]
scores, classes = prob.max(-1), prob.argmax(-1) # column index IS the COCO id
boxes = out["dets"].numpy()[0] # cxcywh, multiply by image W/H
keep = scores > 0.5 # done — no NMS
RF-DETR-Seg (instance segmentation)
rfdetr-seg-{nano,small,medium,large,xlarge,2xlarge}_float32.aimodel
— same
contract plus
masks [1, Q, R/4, R/4]
: per-query FULL-FRAME logit planes at
stride 4 (host: sigmoid > 0.5; no ROI plumbing, no NMS). All six gate on CPU
and GPU with binary-mask IoU 1.000 on stable scenes. M4 Max GPU:
seg-nano 312²
10.7 ms
→ seg-2xlarge 768²
59.1 ms
.
Split deployment (
split/
)
split/rfdetr-{nano,medium}_{backbone,head}.aimodel
separate the pure-ViT
backbone (image → features) from the deformable head (features → dets/labels;
position encodings baked in). The chain is bit-exact vs the monolith. Purpose:
per-stage compute-unit preferences — e.g. backbone on the Neural Engine.
Measured honestly: on iOS 27 beta the runtime still executes the backbone on
the GPU delegate even under
.neuralEngine
preference (identical detection
fingerprint, no ANE-compile pause), so today the monolith on GPU is the
fastest config; the split exists so ANE placement can be adopted the moment
the runtime honors it. Regenerate with
export_rf_detr.py --variant <v> --split
.
Conversion
Exported with
conversion/export_rf_detr.py
from
rfdetr==1.7.1
weights. The port surfaced four Core AI converter/runtime bugs
(float-arg
arange
abort, int64-comparison buffer clobber, GPU-delegate
floor/trunc/ceil = identity, cast-pair cancellation) — each worked around numerically
identically; details and minimal repros in
zoo/rf-detr.md
.
License: Apache-2.0 (upstream RF-DETR code and COCO-pretrained weights are Apache-2.0).
Runs of mlboydaisuke RF-DETR-CoreAI on huggingface.co
144
Total runs
0
24-hour runs
0
3-day runs
2
7-day runs
-37
30-day runs
More Information About RF-DETR-CoreAI huggingface.co Model
Url of RF-DETR-CoreAI
RF-DETR-CoreAI huggingface.co Url
Provider of RF-DETR-CoreAI huggingface.co
Other API from mlboydaisuke
Updated:September 26 2026
Total runs:
1.4K
Run Growth:
336
Growth Rate:
24.69%
Total runs:
648
Run Growth:
7
Growth Rate:
1.08%
Total runs:
497
Run Growth:
77
Growth Rate:
15.49%
Total runs:
394
Run Growth:
347
Growth Rate:
88.07%
Total runs:
381
Run Growth:
-12
Growth Rate:
-3.15%
Total runs:
355
Run Growth:
0
Growth Rate:
0.00%
Total runs:
291
Run Growth:
285
Growth Rate:
97.94%
Total runs:
285
Run Growth:
247
Growth Rate:
86.67%
Total runs:
230
Run Growth:
17
Growth Rate:
7.39%
Total runs:
225
Run Growth:
6
Growth Rate:
2.62%
Total runs:
219
Run Growth:
106
Growth Rate:
51.71%
Total runs:
204
Run Growth:
174
Growth Rate:
85.29%
Total runs:
202
Run Growth:
177
Growth Rate:
88.50%
Total runs:
200
Run Growth:
167
Growth Rate:
83.50%
Total runs:
187
Run Growth:
-60
Growth Rate:
-33.71%
Total runs:
182
Run Growth:
-208
Growth Rate:
-149.64%
Total runs:
181
Run Growth:
5
Growth Rate:
2.39%
Total runs:
171
Run Growth:
122
Growth Rate:
90.37%
Total runs:
169
Run Growth:
156
Growth Rate:
92.31%
Total runs:
145
Run Growth:
67
Growth Rate:
72.04%
Total runs:
139
Run Growth:
-63
Growth Rate:
-41.18%
Total runs:
135
Run Growth:
42
Growth Rate:
31.11%
Total runs:
115
Run Growth:
-177
Growth Rate:
-129.20%
Total runs:
114
Run Growth:
61
Growth Rate:
53.51%
Total runs:
111
Run Growth:
111
Growth Rate:
100.00%
Total runs:
107
Run Growth:
61
Growth Rate:
57.01%
Total runs:
106
Run Growth:
-60
Growth Rate:
-57.14%
Total runs:
105
Run Growth:
48
Growth Rate:
44.44%
Total runs:
103
Run Growth:
58
Growth Rate:
56.31%
Total runs:
103
Run Growth:
38
Growth Rate:
31.15%
Total runs:
101
Run Growth:
-8
Growth Rate:
-7.92%
Total runs:
97
Run Growth:
23
Growth Rate:
23.71%
Total runs:
97
Run Growth:
35
Growth Rate:
35.00%
Total runs:
93
Run Growth:
-175
Growth Rate:
-180.41%
Total runs:
90
Run Growth:
79
Growth Rate:
87.78%
Total runs:
87
Run Growth:
87
Growth Rate:
100.00%
Total runs:
84
Run Growth:
-15
Growth Rate:
-16.30%
Total runs:
81
Run Growth:
-562
Growth Rate:
-693.83%
Total runs:
81
Run Growth:
-102
Growth Rate:
-139.73%
Total runs:
81
Run Growth:
-224
Growth Rate:
-269.88%
Total runs:
81
Run Growth:
-173
Growth Rate:
-210.98%
Total runs:
80
Run Growth:
-53
Growth Rate:
-62.35%
Total runs:
76
Run Growth:
-29
Growth Rate:
-38.16%
Total runs:
74
Run Growth:
-35
Growth Rate:
-46.05%
Total runs:
74
Run Growth:
2
Growth Rate:
2.70%
Total runs:
73
Run Growth:
-144
Growth Rate:
-189.47%
Total runs:
72
Run Growth:
-30
Growth Rate:
-34.09%
Total runs:
71
Run Growth:
9
Growth Rate:
13.85%
Total runs:
69
Run Growth:
44
Growth Rate:
63.77%
Total runs:
68
Run Growth:
-63
Growth Rate:
-94.03%
Total runs:
66
Run Growth:
-6
Growth Rate:
-11.76%
Total runs:
65
Run Growth:
14
Growth Rate:
22.22%
Total runs:
64
Run Growth:
6
Growth Rate:
11.11%
Total runs:
63
Run Growth:
37
Growth Rate:
57.81%
Total runs:
59
Run Growth:
37
Growth Rate:
62.71%
Total runs:
59
Run Growth:
-20
Growth Rate:
-38.46%
Total runs:
59
Run Growth:
2
Growth Rate:
3.92%
Total runs:
58
Run Growth:
-46
Growth Rate:
-75.41%
Total runs:
56
Run Growth:
36
Growth Rate:
64.29%
Total runs:
56
Run Growth:
-14
Growth Rate:
-25.45%
Total runs:
54
Run Growth:
-35
Growth Rate:
-56.45%
Total runs:
54
Run Growth:
-41
Growth Rate:
-71.93%
Total runs:
54
Run Growth:
-27
Growth Rate:
-43.55%
Total runs:
53
Run Growth:
10
Growth Rate:
14.08%
Total runs:
53
Run Growth:
-50
Growth Rate:
-104.17%
Total runs:
52
Run Growth:
-111
Growth Rate:
-241.30%
Total runs:
50
Run Growth:
27
Growth Rate:
51.92%
Total runs:
50
Run Growth:
50
Growth Rate:
100.00%
Total runs:
49
Run Growth:
-1
Growth Rate:
-2.33%
Total runs:
47
Run Growth:
-19
Growth Rate:
-46.34%
Total runs:
46
Run Growth:
-18
Growth Rate:
-42.86%
Total runs:
45
Run Growth:
-124
Growth Rate:
-326.32%
Total runs:
44
Run Growth:
-9
Growth Rate:
-20.93%
Total runs:
44
Run Growth:
-12
Growth Rate:
-27.27%
Total runs:
44
Run Growth:
-54
Growth Rate:
-98.18%
Total runs:
44
Run Growth:
21
Growth Rate:
47.73%
Total runs:
43
Run Growth:
-20
Growth Rate:
-43.48%
Total runs:
43
Run Growth:
-58
Growth Rate:
-134.88%
Total runs:
42
Run Growth:
-37
Growth Rate:
-82.22%
Total runs:
40
Run Growth:
15
Growth Rate:
100.00%
Total runs:
39
Run Growth:
10
Growth Rate:
25.64%
Total runs:
39
Run Growth:
39
Growth Rate:
100.00%
Total runs:
39
Run Growth:
-71
Growth Rate:
-144.90%
Total runs:
39
Run Growth:
8
Growth Rate:
22.22%
Total runs:
38
Run Growth:
-7
Growth Rate:
-22.58%
Total runs:
37
Run Growth:
21
Growth Rate:
56.76%
Total runs:
35
Run Growth:
29
Growth Rate:
82.86%
Total runs:
33
Run Growth:
-19
Growth Rate:
-51.35%
Total runs:
33
Run Growth:
33
Growth Rate:
100.00%
Total runs:
33
Run Growth:
-20
Growth Rate:
-71.43%
Total runs:
33
Run Growth:
-9
Growth Rate:
-21.95%
Total runs:
32
Run Growth:
17
Growth Rate:
53.13%
Total runs:
31
Run Growth:
-82
Growth Rate:
-248.48%
Total runs:
31
Run Growth:
-73
Growth Rate:
-221.21%
Total runs:
30
Run Growth:
-38
Growth Rate:
-111.76%
Total runs:
30
Run Growth:
-8
Growth Rate:
-22.22%
Total runs:
30
Run Growth:
-26
Growth Rate:
-68.42%