Coqueta Studio - local video segmentation

MediaPipe Tasks Vision 0.10.21, Apache License 2.0 (LICENSE.txt).
Source: https://github.com/google-ai-edge/mediapipe/tree/v0.10.21
Pinned runtime: https://www.npmjs.com/package/@mediapipe/tasks-vision/v/0.10.21

Google MediaPipe models:
Selfie segmenter float16 v1:
https://storage.googleapis.com/mediapipe-models/image_segmenter/selfie_segmenter/float16/1/selfie_segmenter.tflite
Magic Touch float32 v1:
https://storage.googleapis.com/mediapipe-models/interactive_segmenter/magic_touch/float32/1/magic_touch.tflite
EfficientDet-Lite2 int8, local snapshot 2026-09-24 (person detector):
https://storage.googleapis.com/mediapipe-models/object_detector/efficientdet_lite2/int8/latest/efficientdet_lite2.tflite
SHA256: B3F50554CB0EA559E90328845F7D9BA4D13C8BFF372914D24E06BC8BB72FA896
Local file: efficientdet-lite2-int8-v1.tflite (v1 is the local snapshot name).
Model information: https://ai.google.dev/edge/mediapipe/solutions/vision/object_detector
Upstream EfficientDet code: https://github.com/google/automl/tree/master/efficientdet (Apache-2.0).

Selfie Multiclass 256x256 float32, Apache-2.0, local snapshot 2026-09-23:
https://storage.googleapis.com/mediapipe-models/image_segmenter/selfie_multiclass_256x256/float32/latest/selfie_multiclass_256x256.tflite
SHA256: C6748B1253A99067EF71F7E26CA71096CD449BAEFA8F101900EA23016507E0E0
Model card: https://storage.googleapis.com/mediapipe-assets/Model%20Card%20Multiclass%20Segmentation.pdf
Model information and applicable terms:
https://ai.google.dev/edge/mediapipe/solutions/vision/image_segmenter
https://ai.google.dev/edge/mediapipe/solutions/vision/interactive_segmenter
https://ai.google.dev/edge/mediapipe/solutions/vision/interactive_segmenter#models

SAM 2.1 hiera tiny (Segment Anything 2), Meta, Apache License 2.0 (LICENSE.txt):
https://github.com/facebookresearch/sam2
ONNX export: https://huggingface.co/onnx-community/sam2.1-hiera-tiny-ONNX
Local files: sam2/sam2-encoder.part*.bin and sam2/sam2-decoder.part1.bin (see sam2/manifest.json).
Change: weights stored as per-channel int8 with DequantizeLinear (computation stays in floating point) and one
positional table stored as float16; the model is split into parts under 25 MB. Used for precise person/object
cut-outs and re-identification after scene changes, entirely on the device.
TEED (Tiny and Efficient Edge Detector), MIT: teed.onnx (248 KB), BIPED/5 weights from
https://github.com/xavysp/TEED (Soria, Li, Rouhani and Sappa, ICCV 2023 Workshops), exported to ONNX unchanged.
Used to snap cut-out edges to the drawn contour lines.
MIT License
Copyright (c) 2022 Xavier Soria Poma
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated
documentation files (the "Software"), to deal in the Software without restriction, including without limitation the
rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit
persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the
Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

MediaBunny 1.56.2: see ../video/NOTICE.txt.
GIF export: gifenc 1.0.3, MIT, Copyright (c) 2017 Matt DesLauriers.
Source: https://github.com/mattdesl/gifenc
Pinned distribution: https://www.npmjs.com/package/gifenc/v/1.0.3
Local module: ../video/gifenc-1.0.3.mjs
Full license: ../video/gifenc-LICENSE.txt
No commercial API is called. Model results are approximate.