project_kelpie/anima_module.yaml at main · RobotFlow-Labs/project_kelpie · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
module:
  name: project_kelpie
  codename: KELPIE
  wave: 10
  theme: WARDOG
  tier: T2
  defense_score: 36

paper:
  title: Prototype-Based Low Altitude UAV Semantic Segmentation
  short_title: PBSeg
  arxiv_id: "2604.01550"
  github: https://github.com/zhangda1018/PBSeg
  verification_status: PARTIAL_VERIFIED_CTO_REVIEW
  date: "2026-04-10"

model:
  task: semantic_segmentation
  architecture: prototype_based_cross_attention
  backbone: resnet50
  num_classes: 8
  num_queries: 100
  hidden_dim: 256
  pbca_layers: 6

compute:
  backends: [cuda, mlx, cpu]
  default_device: auto
  batch_size_cuda: 16
  batch_size_mlx: 4
  batch_size_cpu: 2
  min_vram_gb: 16

datasets:
  primary: internal_1_8m_uav
  supported:
    - uavid
    - udd6
    - visdrone
    - uavdt
    - dronevehicle
    - seadronessee
  shared_root: /Volumes/AIFlowDev/RobotFlowLabs/datasets/

serve:
  entrypoint: anima_kelpie.serve:main
  protocol: http
  port: 8080
  endpoints:
    - GET /health
    - GET /metadata
    - POST /predict
  healthcheck: http://localhost:8080/health
  min_ram_gb: 4
  gpu_required: false
  extras:
    - fastapi
    - uvicorn

training:
  script: python -m anima_kelpie.train
  config: configs/default.toml
  epochs: 100
  lr: 0.001
  optimizer: adamw
  weight_decay: 0.0001
  loss: mask_classification
  val_ratio: 0.1
  output_dir: /mnt/artifacts-datai/checkpoints/project_kelpie
  log_dir: /mnt/artifacts-datai/logs/project_kelpie

export:
  script: python -m anima_kelpie.export
  output_dir: /mnt/artifacts-datai/exports/project_kelpie
  formats:
    - pt
    - safetensors
    - onnx
    - tensorrt_fp16
    - tensorrt_fp32

huggingface:
  repo: ilessio-aiflowlab/project_kelpie
  private: true

integration:
  yolo26_required: true
  ros2: anima-ros2-bridge
  simulator: anima-Ros2-Gazebo

status:
  phase: phase_1_scaffold
  mvp_readiness: 55
  blockers:
    - waiting_for_uavid_udd6_access
    - gpus_busy_with_nighthawk_mega_dataset
    - upstream_repo_empty_reimplemented_from_paper
  last_update: "2026-04-11"