TensorRT YOLO v5.0.0 - Release Notes

Breaking Changes

• Add EfficientRotatedNMS Plugin for OBB Model Inference (0caaf27)
• Add Support for YOLO OBB Export and Inference with EfficientRotatedNMS Plugin (c2686cc)
• Add Support for YOLO11 Detect and OBB Detect (01c8eb4)
• Refactor: Enable plugin compilation as a dynamic library independently of TensorRT-OSS (4506565)
• Add EfficientIdxNMS Plugin for Segmentation Model Inference (72324ed)
• feat(YOLOv10): Update detect head to reduce ONNX size (701e211)
• feat(cli): Removed YOLOv6/7/9 export options and directed to official repositories (af0783b)
• feat(Segment): Add support for exporting YOLOv3 and YOLOv5 segmentation models to ONNX using the EfficientIdxNMS Plugin (efe18ab)
• refactor: Simplify inference with templates, separate OBB, update bindings (5af55b1)
• feat(Segment): Add support YOLO Segmentation inference module with C++ and Python bindings (ee3e2f9)
• build(Dockerfile): Update to use newer base image (7cf37ee)
• feat(Segment): Add example for YOLO Segmentation (135d213)
• feat(Pose): Add support for exporting Ultralytics Pose models to ONNX using the EfficientIdxNMS Plugin (5fedc5f)
• feat(Pose): Add YOLO Pose Estimation inference module with C++ and Python bindings, include example (687b8ea)

Bug Fixes

• Fix TypeError: milliseconds(): incompatible function arguments. (931e8b1)
• Fix: Aligned precision (12ea4da)
• fix: Correct syntax error in xmake.lua (f85339a)
• fix: Fix type mismatch in YOLOv10 detect head (b2900a2)
• fix: Corrected the torch.hub.load function to use _verbose=False instead of verbose=False to properly silence the output. (9f2c52e)
• fix: Address missing imports, dependency issues, and default settings in CLI (05e6355)
• fix: Correct rgbPtr comment from BGR to RGB (9488e10)
• fix: Corrected the description of the cudaMem parameter in the comments (d712141)
• fix: Replace raise with sys.exit(1) for consistent exception handling (9316089)
• fix(#59): Resolved Custom Plugin Error with trtexec for TensorRT versions 10.0.x to 10.1.x (758da3b)
• fix: Ensure output directory creation is conditional on args.output presence (06a577c)
• fix: Resolve metadata inconsistency in onnx-simplifier package installation (1d9a7a4)
• feat(Pose): Adapt code to support 2D and 3D keypoint coordinates (7922d89)

Full Changelog: v4.0...v5.0.0

TensorRT YOLO v4.0 - Release Notes

Breaking Changes

• Add Dockerfile for building project Docker image (73060db)
• Add support for YOLOv3 and Ultralytics model export (8f2af94)
• Added Support for YOLOv10 (62071cb)
• Refactor deploy library (8a4df33)
• Use CUDA Graph to Accelerate Static Model Inference (3576c78)
• Refactor: Add BaseDet base class, refactor DeployDet and DeployCGDet (416e77b)
• Add streaming video support using VideoPipe (#17, #19)
• feat: major update with pybind11 integration and new 4.0 tag (3244eea)

Bug Fixes

• Fix incorrect time interval calculation (a3dee2a)
• Fix: Include <cstring> to resolve "memcpy is not a member of std" error in Linux (1b763d9)
• Update detection output variable names for clarity (fe67b01)
• Add cstdint to resolve Linux compilation issue (f602dc4)
• Fix: Graph input and output tensors must include dtype information. (2cacb7d)

Full Changelog: v3.0...v4.0

TensorRT YOLO v3.0 - Release Notes

Breaking Changes

• Add TensorRT INT8 PTQ support (87f67ff)
• Add C++ inference implementation (0f3069f)
• Implemented parallel preprocessing with multiple streams (86d6175)
• Refactor C++ inference code to support dynamic and static libraries (425a1a4)
• Refactored Python code related to TensorRT-YOLO and packaged it as tensorrt_yolo. (a10ebc8)

Bug Fixes

• Fix batch visualize bug (9125219)
• Remove deleted move constructor and move assignment operator (e287342)
• Fix duplicate imports (1237e21)
• Fix bug (24ea950)

Full Changelog: v2.0...v3.0

TensorRT YOLO v2.0 - Release Notes

Breaking Changes

• Implement YOLOv9 Export to ONNX and TensorRT with EfficientNMS Plugin (249bfab)
• Remove FLOAT16 ONNX export and add support for Dynamic Shape export (9ec1f29)
• Enable dynamic shape inference with CUDA Python and TensorRT 8.6.1 for inference (3286450)

Bug Fixes

• Fix bounding boxes rescales bug (0f90cd0)
• Fix AttributeError in YOLOv9 Model Export (#8)

Full Changelog: v1.0...v2.0

TensorRT YOLO v1.0 - Release Notes

Breaking Changes

• Supports FLOAT32, FLOAT16 ONNX export, and TensorRT inference
• Supports YOLOv5, YOLOv8, PP-YOLOE, and PP-YOLOE+
• Integrates EfficientNMS TensorRT plugin for accelerated post-processing
• Utilizes CUDA kernel functions to accelerate preprocessing
• Supports Python inference

Bug Fixes

• Fix pycuda.driver.CompileError on Jetson (#1)
• Fix Engine Deserialization Failed using YOLOv8 Exported Engine (#2)
• Fix Precision Anomalies in YOLOv8 FP16 Engine (#3)
• Fix YOLOv8 EfficientNMS output shape abnormality (0e542ee)
• Fix trtexec Conversion Failure for YOLOv5 and YOLOv8 ONNX Models on Linux) (#4)
• Fix Inference Anomaly Caused by preprocess.cu on Linux (#5)

Full Changelog: https://github.com/laugh12321/TensorRT-YOLO/commits/v1.0