LaMa (big-lama) — Roman Suvorov et al., Apache License 2.0 (text: LICENSE-Apache-2.0.txt). ONNX export: Carve (LaMa-ONNX); int4/int8 quantization and rank<=4 rewrite: othytenk (lama-int4-int8). Training data attribution: Places2 (MIT CSAIL) — model trained on Places2. ONNX Runtime Web 1.30.0 — Microsoft, MIT License (vendor/ort-webgpu-130/LICENSE.txt). U²-NetP (u2netp.onnx) — Xuebin Qin et al., "U²-Net: Going Deeper with Nested U-Structure for Salient Object Detection", Pattern Recognition 2020; Apache License 2.0. ONNX file as published by the rembg project. SlimSAM-77 (vision_encoder_fp16.onnx, prompt_encoder_mask_decoder_fp16.onnx) — Zigeng Chen et al., "0.1% Data Makes Segment Anything Slim", NeurIPS 2024; Apache License 2.0; derived from Segment Anything (Meta AI, Apache 2.0). ONNX/fp16 conversion: Xenova (Transformers.js).