YOLOv7 Model ONNX

YOLOv7 Model ONNX Export Parameters

This release contains YOLOv7 models exported from the original yolov7.pt and yolov7_qat.pt using the following parameters:

• Grid Export (--grid): Added Detect() Layer
• End-to-End Export (--end2end): Enables end-to-end export. (Efficient NMS)
• Dynamic Batch Export (--dynamic-batch/ dynamic): Uses dynamic batch export.
• ONNX Simplification (--simplify): the ONNX Simplifier helps optimize and enhance the performance of ONNX models.
• Top-k All 100 (--topk-all 100): Keeps the top 100 detections for all classes.
• IOU Threshold 0.65 (--iou-thres 0.65): Sets the IOU (Intersection over Union) threshold to 0.65.
• Confidence Threshold 0.35 (--conf-thres 0.35): Sets the confidence threshold to 0.35.
• Input Image Size 640x640 (--img-size 640 640): Sets the input image size to 640x640 pixels.