v0.51.0

What's Changed

• Add perception encoder workflow blocks by @hansent in #1350

• Feat/stability ai outpainting by @grzegorz-roboflow in #1367

• Respect image parent_id property when deserializing by @brunopicinin in #1363

• Fix zone filter anchor extraction for non-center detection reference points by @ryanjball in #1372

• Fix 'Invalid date error' when creating new workflow by @grzegorz-roboflow in #1365

• Fix velocity block - sv.Detections[data] items must be np.array elements by @grzegorz-roboflow in #1368

• communicate failed pipeline over webrtc by @grzegorz-roboflow in #1376

• Update owlv2.py by @stellasphere in #1375

• docs: Add installation and a quickstart code snippet in the getting-started page in docs by @soumik12345 in #1362

• Change ONVIF Block Name by @lou-roboflow in #1371

New Contributors

• @soumik12345 made their first contribution in #1362

Full Changelog: v0.50.5...v0.51.0

v0.50.5

Onvif sink

Inference users can now control PTZ cameras!
ONVIF block allows a workflow to control an ONVIF capable PTZ camera to follow a detected object.

Block can operate in two modes:

• Follow: The object it follows is the maximum confidence prediction out of all predictions passed into it. To follow
  a specific object, use the appropriate filters on the predictiion object to specify the object you want to
  follow. Additionally if a tracker is used, the camera will follow the tracked object until it disappears.
  Additionally, zoom can be toggled to get the camera to zoom into a position.
  The tracking block uses the ONVIF continuous movement service. Tracking is adjusted on each successive
  workflow execution. If workflow execution stops, and the camera is currently moving, the camera will continue
  moving until it reaches the limits and will no longer be following an object.
• Move to Preset: The camera can also move to a defined preset position. The camera must support the GotoPreset service.

PID tuning is necessary for this block to avoid having the camera overshoot and hunt. Having a
significant lag between the camera movement and video (using a lazy buffer consumption strategy) can make
tuning extremely difficult. Using an eager buffer consumption strategy is recommended. Increasing the dead
zone can also help, but can affect zooming.

Change by @lou-roboflow in #1348

⚡ Speed-ups (with https://github.com/codeflash-ai)

• ⚡️ Speed up function load_blocks by 409% in PR #1348 (ONVIF_Sink) by @codeflash-ai in #1349
• ⚡️ Speed up function receive_socket_data by 16% by @misrasaurabh1 in #1177
• ⚡️ Speed up method BlockManifest.describe_outputs by 1,091% by @misrasaurabh1 in #1181
• ⚡️ 11% (0.11x) speedup for RFDETRObjectDetection.postprocess in inference/models/rfdetr/rfdetr.py by @aseembits93 in #1217

🧑‍🏭 Maintanence

• Bump the pip group across 2 directories with 1 update by @dependabot in #1356
• if DEVICE_ID env is set, store it in resource_details by @grzegorz-roboflow in #1358
• Make model unsupported message more descriptive by @bigbitbus in #1357
• add benchmark configuration and case for codeflash by @KRRT7 in #1254
• Fix/add default fallback to class replacement by @grzegorz-roboflow in #1359

Documentation updates

• Fix typo: prometeus -> prometheus by @iurisilvio in #1353
• remove trailing space in hyperlink text by @afomi in #1346
• fix small typo by @afomi in #1345

🏅 New Contributors

• @aseembits93 made their first contribution in #1217
• @afomi made their first contribution in #1346

Full Changelog: v0.50.4...v0.50.5

v0.50.4

What's Changed

• Introduce OpenAI API key kind by @hansent in #1303
• ⚡️ Speed up function execute_gpt_4v_request by 100% in PR #1214 (openai-apikey-passthrough) by @codeflash-ai in #1280
• API Key Passthrough by @yeldarby in #1316
• Fix owlvit for torch 27 by @probicheaux in #1324
• feat: add PerceptionEncoder model implementation with tests by @probicheaux in #1325
• Revert "feat: add PerceptionEncoder model implementation with tests" by @PawelPeczek-Roboflow in #1329
• Add smolvlm train by @Matvezy in #1327
• Add smolvlm256 by @Matvezy in #1337
• version1 overview by @Azielll in #1333
• include editor.html in app bundle builds by @hansent in #1334
• Add serialisation error logs by @PawelPeczek-Roboflow in #1331
• Add fix for empty segmentations by @PawelPeczek-Roboflow in #1339
• Fix RTMP lack of smoothness by @grzegorz-roboflow in #1328
• Fix/make logs json serializable by @grzegorz-roboflow in #1340
• When dynamic zone was not found, return empty result by @grzegorz-roboflow in #1341
• Revert "Revert "feat: add PerceptionEncoder model implementation with tests"" by @probicheaux in #1338
• Try to fix gpu build by @PawelPeczek-Roboflow in #1342
• Fix default name by @PawelPeczek-Roboflow in #1343
• v3 openAI block with support for api key passthrough by @hansent in #1214
• last by @Azielll in #1344

New Contributors

• @Azielll made their first contribution in #1333

Full Changelog: v0.50.3...v0.50.4

v0.50.3

What's Changed

• Feat/handle segmentation models in detections consensus by @grzegorz-roboflow in #1321

Full Changelog: v0.50.2...v0.50.3

v0.50.2

What's Changed

• allow app bundle actions to write to release by @hansent in #1315
• Fix JSON parser for direct strings by @hansent in #1306
• Bump next from 15.2.4 to 15.3.2 in /inference/landing in the npm_and_yarn group across 1 directory by @dependabot in #1288
• Rfdetr 0 index by @probicheaux in #1318
• fix task name for instance seg api endpoint by @hansent in #1317
• Fix OWLv2 numpy image unload by @balthazur in #1319
• Fix OwlV2 test by @PawelPeczek-Roboflow in #1320

Full Changelog: v0.50.1...v0.50.2

v0.50.1

What's Changed

• Fix dynamic zones least squares fitting defect by @grzegorz-roboflow in #1314

Full Changelog: v0.50.0...v0.50.1

v0.50.0

What's Changed

• Fix documentation typo by @yeldarby in #1295
• Fix stale rich requirement by @yeldarby in #1300
• Add AGENTS.md guide by @yeldarby in #1299
• Improve Workflows docs writing by @yeldarby in #1298
• Fix compatibility typos by @yeldarby in #1297
• Fix single return in get_system_info by @yeldarby in #1296
• Document Codex Environment Config by @yeldarby in #1304
• Add OSX and windows build scripts and github actions by @hansent in #1302
• Recalculate perspective transformation matrix for each frame if extended by @grzegorz-roboflow in #1310
• Extend dynamic zone block with dynamic zone fitting of edges produced by convex hull and approxPoly by @grzegorz-roboflow in #1311
• Fix CodeQL scan by @hansent in #1309
• Improvements to webrtc stability by @grzegorz-roboflow in #1305
• silcence libav swscaler warnings by @grzegorz-roboflow in #1312
• App Bundle fixes for inference repo build by @hansent in #1307
• Accept static reference polygon as input in size measurement block by @grzegorz-roboflow in #1313

Full Changelog: v0.49.5...v0.50.0

v0.49.5

What's Changed

• Move call to build_roboflow_api_headers from payload_helpers to collector by @grzegorz-roboflow in #1290
• Fix collector by @grzegorz-roboflow in #1291
• 0.49.5 by @grzegorz-roboflow in #1293

Full Changelog: v0.49.3...v0.49.5

v0.49.3

What's Changed

• Fixing CLA for gemini models stuff by @abelriboulot in #1286
• When extending perspective correction block, also extend transformed rect dimensions proportionally to original dimensions by @grzegorz-roboflow in #1287
• Added retry to 504 status by @bigbitbus in #1289

Full Changelog: v0.49.2...v0.49.3

v0.49.2

What's Changed

• Added 60s cache expiry for invalid models by @bigbitbus in #1252
• Adding a floor of 100ms for GCP_SERVERLESS by @bigbitbus in #1253
• Fix Moondream2 Workflow long_description. Should be string, not tuple. by @brunopicinin in #1267
• Fix/scale sv detections data polygon by @grzegorz-roboflow in #1268

Full Changelog: v0.49.1...v0.49.2