[Beta Support]: Classification model training failing

[MelvinSnijders]

Describe the problem you are having

When trying to train the model in the classification section, it starts and fails after about 3 seconds.
I'm not sure if this is a bug or a configuration error.

Beta Version

0.17.0-rc1

Issue Category

Object Detection / Detectors

Frigate config file

mqtt:
  user: nas
  password: <REDACTED>
  host: mosquitto

proxy:
  auth_secret: <REDACTED>
  default_role: admin

auth:
  enabled: false

detect:
  enabled: true
  width: 704
  height: 576

detectors:
  coral:
    type: edgetpu
    device: usb

objects:
  track:
    - person
  genai:
    enabled: true

record:
  enabled: true
  continuous:
    days: 7
  motion:
    days: 14
  alerts:
    retain:
      days: 30
      mode: all
  detections:
    retain:
      days: 30
      mode: all

snapshots:
  enabled: true

semantic_search:
  enabled: true
  reindex: false
  model_size: small

genai:
  provider: gemini
  api_key: <REDACTED>
  model: gemini-1.5-flash
go2rtc:
  streams:
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>
    redacted:
      - rtsp://<REDACTED>

cameras:
  redacted:
    ui:
      order: 1
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask: 
        0.213,0.226,0.35,0.247,0.371,0.215,0.372,0.072,1,0.108,1,0,0,0,0,0.385
  redacted:
    ui:
      order: 2
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask: 
        0.47,0.094,0.559,0.094,0.551,0.199,0.805,0.197,0.8,0.071,0.857,0.063,1,0.159,1,0,0.455,0
      threshold: 30
      contour_area: 10
      improve_contrast: true
  redacted:
    ui:
      order: 3
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask:
        - 450,0,294,0,292,81,452,84
  redacted:
    ui:
      order: 4
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask: 0.87,0,1,0,1,0.322,0.91,0.242,0.882,0.186,0.91,0.041
  redacted:
    ui:
      order: 5
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask: 0,0.142,0,0,0.429,0,0.413,0.199,0.247,0.197,0.237,0.139,0.209,0.136,0.144,0.172,0.132,0.108
  redacted:
    ui:
      order: 6
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect
    motion:
      mask:
        - 0.122,0.609,0.447,0.538,0.459,0.115,0.083,0.168
        - 0.498,0.113,0.65,0.097,0.616,0.44,0.482,0.468
  redacted:
    ui:
      order: 7
    ffmpeg:
      inputs:
        - path: rtsp://<REDACTED>
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: 
            rtsp://<REDACTED>
          roles:
            - detect

version: 0.17-0
face_recognition:
  enabled: true
  model_size: small
lpr:
  enabled: false
classification:
  bird:
    enabled: false
  custom:
    Hek:
      enabled: true
      name: Hek
      threshold: 0.8
      state_config:
        cameras:
          hek:
            crop:
              - 0
              - 0.36969779197837066
              - 0.4375184704929708
              - 0.9044425892475573
        motion: true

Relevant Frigate log output

2026-02-14 21:44:53.216344962  10.10.10.105 - - [14/Feb/2026:21:44:53 +0000] "POST /api/classification/Hek/train HTTP/1.1" 200 67 "https://redacted.com/classification" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/145.0.0.0 Safari/537.36" "10.0.0.112" request_time="0.129" upstream_response_time="0.075"
2026-02-14 21:45:08.403228860  Fatal Python error: Illegal instruction
2026-02-14 21:45:08.403231926  
2026-02-14 21:45:08.403535191  Thread 0x00007dee231ff6c0 (most recent call first):
2026-02-14 21:45:08.403696998    File "/opt/frigate/frigate/log.py", line 218 in reader
2026-02-14 21:45:08.403800354    File "/usr/lib/python3.11/threading.py", line 975 in run
2026-02-14 21:45:08.403938065    File "/usr/lib/python3.11/threading.py", line 1038 in _bootstrap_inner
2026-02-14 21:45:08.404036171    File "/usr/lib/python3.11/threading.py", line 995 in _bootstrap
2026-02-14 21:45:08.404095915  
2026-02-14 21:45:08.404174885  Current thread 0x00007dee26a55040 (most recent call first):
2026-02-14 21:45:08.404312044    File "<frozen importlib._bootstrap>", line 241 in _call_with_frames_removed
2026-02-14 21:45:08.404441850    File "<frozen importlib._bootstrap_external>", line 1233 in create_module
2026-02-14 21:45:08.404570615    File "<frozen importlib._bootstrap>", line 573 in module_from_spec
2026-02-14 21:45:08.404666667    File "<frozen importlib._bootstrap>", line 676 in _load_unlocked
2026-02-14 21:45:08.404786965    File "<frozen importlib._bootstrap>", line 1149 in _find_and_load_unlocked
2026-02-14 21:45:08.404886544    File "<frozen importlib._bootstrap>", line 1178 in _find_and_load
2026-02-14 21:45:08.405005679    File "<frozen importlib._bootstrap>", line 241 in _call_with_frames_removed
2026-02-14 21:45:08.405101502    File "<frozen importlib._bootstrap>", line 1234 in _handle_fromlist
2026-02-14 21:45:08.408276402    File "/usr/local/lib/python3.11/dist-packages/tensorflow/python/platform/self_check.py", line 63 in preload_check
2026-02-14 21:45:08.408459831    File "/usr/local/lib/python3.11/dist-packages/tensorflow/python/pywrap_tensorflow.py", line 37 in <module>
2026-02-14 21:45:08.414737907    File "<frozen importlib._bootstrap>", line 241 in _call_with_frames_removed
2026-02-14 21:45:08.414985026    File "<frozen importlib._bootstrap_external>", line 940 in exec_module
2026-02-14 21:45:08.415123098    File "<frozen importlib._bootstrap>", line 690 in _load_unlocked
2026-02-14 21:45:08.419864319    File "<frozen importlib._bootstrap>", line 1149 in _find_and_load_unlocked
2026-02-14 21:45:08.419923651    File "<frozen importlib._bootstrap>", line 1178 in _find_and_load
2026-02-14 21:45:08.419978586    File "<frozen importlib._bootstrap>", line 241 in _call_with_frames_removed
2026-02-14 21:45:08.420033680    File "<frozen importlib._bootstrap>", line 1234 in _handle_fromlist
2026-02-14 21:45:08.420089446    File "/usr/local/lib/python3.11/dist-packages/tensorflow/__init__.py", line 40 in <module>
2026-02-14 21:45:08.420991727    File "<frozen importlib._bootstrap>", line 241 in _call_with_frames_removed
2026-02-14 21:47:14.794457627  
2026-02-14 21:47:14.797199797  Extension modules: numpy.core._multiarray_umath, numpy.core._multiarray_tests, numpy.linalg._umath_linalg, numpy.fft._pocketfft_internal, numpy.random._common, numpy.random.bit_generator, numpy.random._bounded_integers, numpy.random._mt19937, numpy.random.mtrand, numpy.random._philox, numpy.random._pcg64, numpy.random._sfc64, numpy.random._generator, pysqlite3._sqlite3, zmq.backend.cython._zmq, ruamel.yaml.clib._ruamel_yaml, _ruamel_yaml, charset_normalizer.md, requests.packages.charset_normalizer.md, requests.packages.chardet.md, PIL._imaging, playhouse._sqlite_ext, psutil._psutil_linux, scipy._lib._ccallback_c, scipy.ndimage._nd_image, scipy.ndimage._rank_filter_1d, scipy.special._ufuncs_cxx, scipy.special._ellip_harm_2, scipy.special._special_ufuncs, scipy.special._gufuncs, scipy.special._ufuncs, scipy.special._specfun, scipy.special._comb, scipy.linalg._fblas, scipy.linalg._flapack, _cyutility, scipy._cyutility, scipy.linalg.cython_lapack, scipy.linalg._cythonized_array_utils, scipy.linalg._solve_toeplitz, scipy.linalg._decomp_lu_cython, scipy.linalg._matfuncs_schur_sqrtm, scipy.linalg._matfuncs_expm, scipy.linalg._linalg_pythran, scipy.linalg.cython_blas, scipy.linalg._decomp_update, scipy.sparse._sparsetools, _csparsetools, scipy.sparse._csparsetools, _ni_label, scipy.ndimage._ni_label, setproctitle._setproctitle, scipy.spatial._ckdtree, scipy._lib.messagestream, scipy.spatial._qhull, scipy.spatial._voronoi, scipy.spatial._hausdorff, scipy.spatial._distance_wrap, scipy.spatial.transform._rotation, scipy.spatial.transform._rigid_transform, scipy.sparse.linalg._dsolve._superlu, scipy.sparse.linalg._eigen.arpack._arpack, scipy.sparse.linalg._propack._spropack, scipy.sparse.linalg._propack._dpropack, scipy.sparse.linalg._propack._cpropack, scipy.sparse.linalg._propack._zpropack, scipy.optimize._group_columns, scipy.optimize._trlib._trlib, scipy.optimize._lbfgsb, _moduleTNC, scipy.optimize._moduleTNC, scipy.optimize._slsqplib, scipy.optimize._minpack, scipy.optimize._lsq.givens_elimination, scipy.optimize._zeros, scipy._lib._uarray._uarray, scipy.linalg._decomp_interpolative, scipy.optimize._bglu_dense, scipy.optimize._lsap, scipy.optimize._direct, scipy.integrate._odepack, scipy.integrate._quadpack, scipy.integrate._vode, scipy.integrate._dop, scipy.integrate._lsoda, scipy.interpolate._fitpack, scipy.interpolate._dfitpack, scipy.interpolate._dierckx, scipy.interpolate._ppoly, scipy.interpolate._interpnd, scipy.interpolate._rbfinterp_pythran, scipy.interpolate._rgi_cython, scipy.special.cython_special, scipy.stats._stats, scipy.stats._biasedurn, scipy.stats._stats_pythran, scipy.stats._levy_stable.levyst, scipy.stats._ansari_swilk_statistics, scipy.sparse.csgraph._tools, scipy.sparse.csgraph._shortest_path, scipy.sparse.csgraph._traversal, scipy.sparse.csgraph._min_spanning_tree, scipy.sparse.csgraph._flow, scipy.sparse.csgraph._matching, scipy.sparse.csgraph._reordering, scipy.stats._sobol, scipy.stats._qmc_cy, scipy.stats._rcont.rcont, scipy.stats._qmvnt_cy, _cffi_backend, multidict._multidict, yarl._quoting_c, propcache._helpers_c, aiohttp._http_writer, aiohttp._http_parser, aiohttp._websocket.mask, aiohttp._websocket.reader_c, frozenlist._frozenlist, regex._regex, lxml._elementpath, lxml.etree, lxml.builder, ciso8601 (total: 123)
2026-02-14 21:47:15.274884019  [2026-02-14 21:47:15] frigate.util.classification    ERROR   : Training subprocess failed for Hek (exit code: -4)

Relevant go2rtc log output (if applicable)

No response

Install method

Docker Compose

docker-compose file or Docker CLI command

version: "3.9"
services:
  frigate:
    image: ghcr.io/blakeblackshear/frigate:0.17.0-rc1
    restart: unless-stopped
    shm_size: 130mb
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /appdata/frigate:/config
      - /recordings:/media/frigate
      - type: tmpfs
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - 5000:5000 # Needed for homeassistant
      - 8971:8971
      - 8554:8554
    env_file:
      - stack.env
    devices:
      - /dev/bus/usb:/dev/bus/usb

Operating system

Proxmox

CPU / GPU / Hardware

Google Coral

Screenshots

Steps to reproduce

1. Go to "Classification"
2. Create a new state classification
3. Go through the wizard
4. Click on train model
5. Training fails and logs errors in the console

Any other information that may be helpful

No response

[hawkeye217]

See the FAQ, Upgrade issues question 3.

[MelvinSnijders]

Whoops, totally missed that. Changing the CPU type to host fixed it for me. Thanks!

[Vincefhvrf]

Hello, I got he same issue but in home assistant with frigate as an integration, I have no possibility to setup the CPU type to host. Do I miss something?

[NickM-27]

What is your CPU?

[Vincefhvrf]

Hi, my CPU is an Intel, NUC95.
l have tried with another camera, and it works well. I don’t figure out the issue. On 6 , 3 are working, all camera are the same. It fails always when generating. When I’m applying the setting in the config, I can’t see any new snapshot, and nothing related to in the debug.