Download NOTICE from CondadosAI/uniformer_s_coco_pose: direct link, hf CLI and curl.
- Browser
- Download file 3.06 kB
-
https://huggingface.co/CondadosAI/uniformer_s_coco_pose/resolve/main/NOTICE
- Command line
-
hf download hf://CondadosAI/uniformer_s_coco_pose/NOTICE
-
curl -L -o NOTICE https://huggingface.co/CondadosAI/uniformer_s_coco_pose/resolve/main/NOTICE
3.06 kB
| UniFormer-S COCO top-down pose - acaua mirror (pure-PyTorch port) | |
| ================================================================== | |
| This product includes: | |
| 1. PORTED SOURCE CODE: a pure-PyTorch port of the UniFormer-pose | |
| architecture (located in the acaua repository at | |
| src/acaua/adapters/uniformer/pose/) is a derivative work of: | |
| - Sense-X/UniFormer (Apache-2.0) | |
| https://github.com/Sense-X/UniFormer @ main | |
| Files derived from: | |
| pose_estimation/mmpose/models/backbones/uniformer.py | |
| (the global-attention UniFormer backbone, parameter-tree | |
| identical to the detection variant — we reuse acaua's | |
| UniFormer2DDense with hybrid=False, windows=False) | |
| pose_estimation/mmpose/models/keypoint_heads/top_down_simple_head.py | |
| (TopDownSimpleHead: 3x ConvTranspose2d-stride-2 + BN + ReLU | |
| + 1x1 final conv) | |
| pose_estimation/mmpose/core/evaluation/top_down_eval.py | |
| (keypoints_from_heatmaps with post_process='default' branch: | |
| argmax + sign(neighbor_diff) * 0.25 shift) | |
| 2. CONVERTED WEIGHTS: the model.safetensors file in this mirror is a | |
| key-remapped conversion of the upstream pretrained checkpoint: | |
| - upstream repo: Sense-X/UniFormer (Apache-2.0) | |
| - upstream weights: Google Drive file id | |
| 162R0JuTpf3gpLe1IK6oxRoQK7JSj4ylx | |
| - upstream filename: top_down_256x192_global_small.pth | |
| - upstream SHA256: d77059e3e9322c0e20dc89dc0cf2a583ffe2ced7d3e9b350233738add570bc30 | |
| - upstream paper: Li et al., "UniFormer: Unifying Convolution | |
| and Self-attention for Visual Recognition", | |
| arXiv:2201.09450 (ICLR 2022) | |
| Key remap rule (eager, at conversion time): | |
| backbone.* -> backbone.* (no change) | |
| keypoint_head.* -> head.* | |
| Conversion was performed by scripts/convert_uniformer_pose.py in the | |
| acaua repository. The adapter loads the safetensors file under | |
| load_state_dict(strict=True) — key-naming drift between upstream and | |
| the port is a hard error at load time. | |
| 3. BUNDLED DETECTOR: this mirror's adapter loads | |
| CondadosAI/rtmdet_t_coco at adapter-init time so callers get the | |
| full top-down pipeline (detect persons -> per-person crop -> pose -> | |
| inverse-warp) from a single predict(image) call. RTMDet-tiny carries | |
| its own attribution chain — see CondadosAI/rtmdet_t_coco's NOTICE | |
| file. | |
| Mirrored on 2026-04-25 by CondadosAI. | |
| License | |
| ------- | |
| Licensed under the Apache License, Version 2.0 (the "License"); | |
| you may not use this file except in compliance with the License. | |
| You may obtain a copy of the License at | |
| http://www.apache.org/licenses/LICENSE-2.0 | |
| Unless required by applicable law or agreed to in writing, software | |
| distributed under the License is distributed on an "AS IS" BASIS, | |
| WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
| See the License for the specific language governing permissions and | |
| limitations under the License. | |