pose estimation benchmarks on edge AI hardware
Each table compares only results with the same model-size class and batch size. Search this workload · JSON.
Fastest in the best-covered like-for-like group (mid-size CNN, batch 1): Qualcomm Snapdragon X2 Elite (CRD) at 1855 FPS D 2026 (year only). 10 groups, 14 platforms.
Show data — 10 like-for-like tables
mid-size CNN · batch 1 (4 platforms)
✓ Like-for-like: model-size class mid-size CNN, batch 1. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Qualcomm Snapdragon X2 Elite (CRD) | 1855 FPS inference_time · b=1 | hrnet_pose w8a8 · Qualcomm AI Hub onnx (QAIRT 2.50.0.260828221209) · mid-size CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Snapdragon X Elite (CRD) | 938 FPS inference_time · b=1 | hrnet_pose w8a8 · Qualcomm AI Hub onnx (QAIRT 2.50.0.260828221209) · mid-size CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Dragonwing IQ-9075 EVK | 864 FPS inference_time · b=1 | hrnet_pose w8a8 · Qualcomm AI Hub tflite (QAIRT 2.50.0.260828221209) · mid-size CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Dragonwing RB3 Gen 2 (QCS6490) | 291 FPS inference_time · b=1 | hrnet_pose w8a8 · Qualcomm AI Hub onnx (QAIRT 2.50.0.260828221209) · mid-size CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
mobile-class CNN · batch 1 (4 platforms)
✓ Like-for-like: model-size class mobile-class CNN, batch 1. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Qualcomm Snapdragon X2 Elite (CRD) | 6211 FPS inference_time · b=1 | mediapipe_pose / pose_landmark_detector w8a8 · Qualcomm AI Hub precompiled_qnn_onnx (QAIRT 2.50.0.260828221209) · mobile-class CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Snapdragon X Elite (CRD) | 3390 FPS inference_time · b=1 | mediapipe_pose / pose_landmark_detector w8a8 · Qualcomm AI Hub precompiled_qnn_onnx (QAIRT 2.50.0.260828221209) · mobile-class CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Dragonwing IQ-9075 EVK | 2532 FPS inference_time · b=1 | mediapipe_pose / pose_landmark_detector w8a8 · Qualcomm AI Hub qnn_context_binary (QAIRT 2.50.0.260828221209) · mobile-class CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
| Qualcomm Dragonwing RB3 Gen 2 (QCS6490) | 947 FPS inference_time · b=1 | mediapipe_pose / pose_landmark_detector w8a8 · Qualcomm AI Hub qnn_context_binary (QAIRT 2.50.0.260828221209) · mobile-class CNN | D | unknown power n/a no power published | Qualcomm AI Hub Models perf.yaml (BSD-3) 2026 (year only) |
medium (YOLO-m) · batch 1 (3 platforms)
✓ Like-for-like: model-size class medium (YOLO-m), batch 1. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-10H (M.2/PCIe, x86 host) | 79.20 FPS throughput · b=1 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8 (M.2/PCIe, x86 host) | 68.60 FPS throughput · b=1 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 48.50 FPS throughput · b=1 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
medium (YOLO-m) · batch 8 (3 platforms)
✓ Like-for-like: model-size class medium (YOLO-m), batch 8. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-8 (M.2/PCIe, x86 host) | 145 FPS throughput · b=8 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-10H (M.2/PCIe, x86 host) | 129 FPS throughput · b=8 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 81.90 FPS throughput · b=8 | yolov8m_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · medium (YOLO-m) | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
small (YOLO-s) · batch 1 (3 platforms)
✓ Like-for-like: model-size class small (YOLO-s), batch 1. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-8 (M.2/PCIe, x86 host) | 393 FPS throughput · b=1 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · small (YOLO-s) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-10H (M.2/PCIe, x86 host) | 161 FPS throughput · b=1 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · small (YOLO-s) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 107 FPS throughput · b=1 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · small (YOLO-s) | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
small (YOLO-s) · batch 8 (3 platforms)
✓ Like-for-like: model-size class small (YOLO-s), batch 8. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-8 (M.2/PCIe, x86 host) | 393 FPS throughput · b=8 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · small (YOLO-s) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-10H (M.2/PCIe, x86 host) | 217 FPS throughput · b=8 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · small (YOLO-s) | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 202 FPS throughput · b=8 | yolov8s_pose INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · small (YOLO-s) | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
nano (YOLO-n) · batch 1 (1 platforms)
✓ Like-for-like: model-size class nano (YOLO-n), batch 1. Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Apple M4 (10-core GPU, 120 GB/s) | 370 FPS latency_e2e · b=1 | YOLO26n-pose FP16 · Ultralytics 8.4.168 Core AI (coreai-torch 0.4.3), macOS 27.0 · nano (YOLO-n) | B | unknown power n/a no power published | Ultralytics docs: Apple Core AI integration benchmark 2026 (year only) |
unknown · batch 1 (3 platforms)
✓ Like-for-like: model-size class unknown, batch 1 (size or batch not stated by source — unverified). Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-8 (M.2/PCIe, x86 host) | 162 FPS throughput · b=1 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · unknown | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-10H (M.2/PCIe, x86 host) | 125 FPS throughput · b=1 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · unknown | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 85.00 FPS throughput · b=1 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · unknown | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
unknown · batch 8 (3 platforms)
✓ Like-for-like: model-size class unknown, batch 8 (size or batch not stated by source — unverified). Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| Hailo-10H (M.2/PCIe, x86 host) | 184 FPS throughput · b=8 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v5.4.0 / HailoRT (model zoo) · unknown | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-10H) undated |
| Hailo-8 (M.2/PCIe, x86 host) | 162 FPS throughput · b=8 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · unknown | D | 2.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8) undated |
| Hailo-8L (M.2/PCIe, x86 host) | 133 FPS throughput · b=8 | centerpose_regnetx_800mf INT8 (Hailo quantized) · Hailo DFC v2.20.0 / HailoRT (model zoo) · unknown | D | 1.50 W spec reported in result source — spec | Hailo Model Zoo public models (Hailo-8L) undated |
unknown · batch unknown (6 platforms)
✓ Like-for-like: model-size class unknown, batch unknown (size or batch not stated by source — unverified). Model, precision and stack may still differ.
| Platform | Result | Model · precision · stack | Grade | Power | Source |
|---|---|---|---|---|---|
| AMD Alveo V70 (Versal AI Core, XDNA AIE-ML) | 8327 FPS E2E throughput, multi thread · b=unknown | pt_movenet_3.5 INT8 · Vitis AI 3.5 (VART) · unknown | D | 75.00 W spec reported in result source — spec | Vitis AI 3.5 Model Zoo performance table Older data (2023) |
| AMD Versal AI Core VCK190 (VC1902) | 437 FPS E2E throughput, multi thread · b=unknown | pt_movenet_coco_192_192_0.5G INT8 · Vitis AI 3.0 (VART) · unknown | D | unknown power n/a no power published | Vitis AI 3.0 Model Zoo performance table Older data (2023) |
| AMD Versal AI Edge VEK280 (VE2802) | 428 FPS E2E throughput, multi thread · b=unknown | pt_movenet_3.5 INT8 · Vitis AI 3.5 (VART) · unknown | D | unknown power n/a no power published | Vitis AI 3.5 Model Zoo performance table Older data (2023) |
| AMD Zynq UltraScale+ ZCU102 (ZU9EG) | 391 FPS E2E throughput, multi thread · b=unknown | pt_movenet_coco_192_192_0.5G INT8 · Vitis AI 3.0 (VART) · unknown | D | unknown power n/a no power published | Vitis AI 3.0 Model Zoo performance table Older data (2023) |
| AMD Zynq UltraScale+ ZCU104 (ZU7EV) | 372 FPS E2E throughput, multi thread · b=unknown | pt_movenet_coco_192_192_0.5G INT8 · Vitis AI 3.0 (VART) · unknown | D | unknown power n/a no power published | Vitis AI 3.0 Model Zoo performance table Older data (2023) |
| AMD Kria KV260 Vision AI Starter Kit (K26 SOM) | 352 FPS E2E throughput, multi thread · b=unknown | pt_movenet_coco_192_192_0.5G INT8 · Vitis AI 3.0 (VART) · unknown | D | 7.50 W spec reported in result source — spec | Vitis AI 3.0 Model Zoo performance table Older data (2023) |
How to read these numbers
Every number is labelled. Confidence: A MLPerf closed, audited · B documented third-party method · C community / review · D vendor-published, not independently verified · E estimate. Power basis: measured with the result · spec vendor spec/envelope · power n/a not published.
Machine-readable: /find.json (see llms.txt); full tables are in “Show data”. Confidence grades: A audited MLPerf closed · B documented third-party · C community · D vendor-published · E estimate. Unknown values are shown as unknown, never estimated. Generated 2026-10-16 from public sources.