initial commit

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+weights filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,22 @@
+import gradio as gr
+
+from detection_pipeline import DetectionModel
+
+model = DetectionModel()
+
+def predict(image, threshold):
+    preds = model(image)
+    preds = list(filter(lambda x: x[4] > threshold, preds))
+
+    output = model.visualize(image, preds)
+    return output
+
+
+
+interface = gr.Interface(
+    fn=predict,
+    inputs=[gr.Image(label="Input"), gr.Slider(0, 1, .3, step=.05, label="Threshold")],
+    outputs=gr.Image(label="Output")
+)
+
+interface.launch()

args.yaml

@@ -0,0 +1,107 @@
+task: detect
+mode: train
+model: /content/weights/yolov10n.pt
+data: /content/data.yaml
+epochs: 20
+time: null
+patience: 100
+batch: 64
+imgsz: 640
+save: true
+save_period: -1
+val_period: 1
+cache: false
+device: null
+workers: 8
+project: null
+name: train
+exist_ok: false
+pretrained: true
+optimizer: auto
+verbose: true
+seed: 0
+deterministic: true
+single_cls: false
+rect: false
+cos_lr: false
+close_mosaic: 10
+resume: false
+amp: true
+fraction: 1.0
+profile: false
+freeze: null
+multi_scale: false
+overlap_mask: true
+mask_ratio: 4
+dropout: 0.0
+val: true
+split: val
+save_json: false
+save_hybrid: false
+conf: null
+iou: 0.7
+max_det: 300
+half: false
+dnn: false
+plots: true
+source: null
+vid_stride: 1
+stream_buffer: false
+visualize: false
+augment: false
+agnostic_nms: false
+classes: null
+retina_masks: false
+embed: null
+show: false
+save_frames: false
+save_txt: false
+save_conf: false
+save_crop: false
+show_labels: true
+show_conf: true
+show_boxes: true
+line_width: null
+format: torchscript
+keras: false
+optimize: false
+int8: false
+dynamic: false
+simplify: false
+opset: null
+workspace: 4
+nms: false
+lr0: 0.01
+lrf: 0.01
+momentum: 0.937
+weight_decay: 0.0005
+warmup_epochs: 3.0
+warmup_momentum: 0.8
+warmup_bias_lr: 0.1
+box: 7.5
+cls: 0.5
+dfl: 1.5
+pose: 12.0
+kobj: 1.0
+label_smoothing: 0.0
+nbs: 64
+hsv_h: 0.015
+hsv_s: 0.7
+hsv_v: 0.4
+degrees: 0.0
+translate: 0.1
+scale: 0.5
+shear: 0.0
+perspective: 0.0
+flipud: 0.0
+fliplr: 0.5
+bgr: 0.0
+mosaic: 1.0
+mixup: 0.0
+copy_paste: 0.0
+auto_augment: randaugment
+erasing: 0.4
+crop_fraction: 1.0
+cfg: null
+tracker: botsort.yaml
+save_dir: runs/detect/train

data.yaml

@@ -0,0 +1,25 @@
+names:
+- Fall-Detected
+- Gloves
+- Goggles
+- Hardhat
+- Ladder
+- Mask
+- NO-Gloves
+- NO-Goggles
+- NO-Hardhat
+- NO-Mask
+- NO-Safety Vest
+- Person
+- Safety Cone
+- Safety Vest
+nc: 14
+roboflow:
+  license: CC BY 4.0
+  project: personal-protective-equipment-combined-model
+  url: https://universe.roboflow.com/roboflow-universe-projects/personal-protective-equipment-combined-model/dataset/4
+  version: 4
+  workspace: roboflow-universe-projects
+test: ../test/images
+train: Personal-Protective-Equipment---Combined-Model-4/train/images
+val: Personal-Protective-Equipment---Combined-Model-4/valid/images

detection_pipeline.py

@@ -0,0 +1,112 @@
+import onnx
+import onnxruntime as ort
+import numpy as np
+
+import cv2
+import yaml
+
+import copy
+
+class DetectionModel():
+    def __init__(self, model_path="weights/best.onnx"):
+        self.current_input = None
+        self.latest_output = None
+        self.model = None
+        self.model_ckpt = model_path
+
+        if self.__check_model():
+            self.model = ort.InferenceSession(self.model_ckpt)
+        else:
+            raise Exception("Model couldn't be validated using ONNX, please check the checkpoint")
+
+        self._load_labels()
+
+    def __check_model(self):
+        model = onnx.load(self.model_ckpt)
+        try:
+            onnx.checker.check_model(model)
+            return True
+        except:
+            return False
+
+    def _preprepocess_input(self, image: np.ndarray) -> np.ndarray:
+        """ Preprocess the input image
+         
+        Resizes the image to 640x640, transposes the matrix so that it's CxHxW and normalizes the image.
+        Then the result is converted to `np.float32` and returned with the extra `batch` dimension
+
+        Args:
+            image (np.ndarray): The input image
+        
+        Returns:
+            processed_image (np.ndarray): The preprocessed image as 1x3x640x640 `np.float32` array
+
+        """
+        processed_image = copy.deepcopy(image)
+        processed_image = cv2.resize(processed_image, (640, 640))
+        processed_image = processed_image.transpose(2, 0, 1)
+        processed_image = (processed_image / 255.0).astype(np.float32)
+        processed_image = np.expand_dims(processed_image, axis=0)
+
+        return processed_image
+
+
+    def _postprocess_output(self, predictions) -> np.ndarray:
+        """ Postprocess the output of the model
+         
+        Args:
+            predictions (np.ndarray): The output of the model as a `np.ndarray`
+        
+        Returns:
+            detections (np.ndarray): The detections as a `np.ndarray` with shape (N, 6) where N is the number of detections.
+                                      The columns are as follows: [x1, y1, x2, y2, confidence, class]
+
+        """
+        w_ratio = self.current_input.shape[1] / 640
+        h_ratio = self.current_input.shape[0] / 640
+
+        detections = []
+        for pred in predictions:
+            # detections.append([int(pred[0]), int(pred[1]), int(pred[2]), int(pred[3]), pred[4], self.ix2l[pred[5]]])
+            detections.append([int(pred[0] * w_ratio), int(pred[1] * h_ratio), int(pred[2] * w_ratio), int(pred[3] * h_ratio), pred[4], self.ix2l[pred[5]]])
+    
+        return list(detections)
+
+    def _load_labels(self):
+        with open("data.yaml", "r") as f:
+            data = yaml.safe_load(f)
+        
+        self.labels = data['names']
+        self.l2ix = {l:i for i, l in enumerate(self.labels)}
+        self.ix2l = {i:l for i, l in enumerate(self.labels)}
+
+    def __call__(self, image: np.ndarray):
+
+        processed_image = self._preprepocess_input(image)
+
+        self.latest_output = list(self.model.run(None, {"images": processed_image})[0][0])
+        self.current_input = image
+
+        detections = self._postprocess_output(self.latest_output)
+
+        return detections
+
+    def visualize(self, input_image: np.ndarray, detections: list[list]) -> np.ndarray:
+        """ Visualizes the detections on the current input image
+        
+        Args:
+            detections (list[list]): The detections as a list of lists
+        
+        Returns:
+            image (np.ndarray): The image with the detections drawn on it
+
+        """
+        image = copy.deepcopy(input_image)
+        for det in detections:
+            x1, y1, x2, y2, conf, label = det
+            cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)
+            cv2.putText(image, f"{label}: {conf:.3f}", (x1, y1), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 0, 0), 2, cv2.LINE_AA)
+        
+        image = cv2.resize(image, self.current_input.shape[:2][::-1])
+
+        return image

events.out.tfevents.1716932751.3799194233d7.8657.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68b83913446010168055f71af86af8eff75de78eb49059d17819384f96a73829
+size 19192

pyproject.toml

@@ -0,0 +1,30 @@
+[tool.poetry]
+name = "ppe-detector-yolov10"
+version = "0.1.0"
+description = ""
+authors = ["Alper Balbay <[email protected]>"]
+readme = "README.md"
+
+[tool.poetry.dependencies]
+python = "^3.11"
+onnxruntime = "^1.18.1"
+gradio = "^4.37.2"
+onnx = "^1.16.1"
+opencv-contrib-python = "^4.10.0.84"
+pillow = "^10.4.0"
+supervision = "^0.21.0"
+ultralytics = "^8.2.48"
+pyyaml = "^6.0.1"
+
+
+[tool.poetry.group.dev.dependencies]
+pytest = "^8.2.2"
+ipython = "^8.26.0"
+ipykernel = "^6.29.5"
+jupyter = "^1.0.0"
+ruff = "^0.5.0"
+pre-commit = "^3.7.1"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

results/results.csv

@@ -0,0 +1,21 @@
+                  epoch,           train/box_om,           train/cls_om,           train/dfl_om,           train/box_oo,           train/cls_oo,           train/dfl_oo,   metrics/precision(B),      metrics/recall(B),       metrics/mAP50(B),    metrics/mAP50-95(B),             val/box_om,             val/cls_om,             val/dfl_om,             val/box_oo,             val/cls_oo,             val/dfl_oo,                 lr/pg0,                 lr/pg1,                 lr/pg2
+                      1,                 1.6175,                 2.7954,                 1.4461,                 1.5534,                 5.1155,                 1.2799,                0.48128,                0.42329,                0.39035,                0.22623,                 1.6007,                 2.3219,                 1.4647,                 1.5281,                 5.6513,                 1.3244,             0.00018495,             0.00018495,             0.00018495
+                      2,                 1.5467,                  1.791,                 1.3574,                 1.5459,                 2.8324,                 1.2449,                0.50376,                0.46524,                0.42232,                 0.2272,                 1.6174,                 1.9051,                 1.4696,                 1.6003,                 2.6558,                 1.3655,             0.00035195,             0.00035195,             0.00035195
+                      3,                 1.5489,                 1.5434,                 1.3614,                 1.5918,                 1.9494,                 1.2795,                0.56758,                 0.5093,                0.49855,                0.26267,                  1.611,                 1.5642,                 1.4715,                 1.6232,                 1.8324,                 1.4041,             0.00050061,             0.00050061,             0.00050061
+                      4,                 1.5166,                 1.4101,                 1.3394,                 1.5802,                 1.7031,                 1.2834,                0.62729,                0.54169,                0.53358,                  0.292,                 1.6274,                 1.4076,                 1.5022,                 1.6425,                 1.6105,                 1.4478,             0.00047343,             0.00047343,             0.00047343
+                      5,                 1.4889,                 1.3234,                 1.3191,                 1.5672,                   1.57,                 1.2735,                0.62211,                0.59476,                 0.6077,                0.34404,                 1.5323,                  1.262,                 1.4028,                 1.5889,                 1.3627,                 1.3654,             0.00047343,             0.00047343,             0.00047343
+                      6,                 1.4626,                 1.2463,                 1.2983,                   1.55,                 1.4592,                 1.2602,                0.66122,                0.61622,                0.62095,                0.36023,                  1.513,                 1.2064,                 1.3778,                 1.5598,                 1.3171,                 1.3557,             0.00044591,             0.00044591,             0.00044591
+                      7,                 1.4298,                 1.1899,                 1.2781,                 1.5266,                 1.3888,                 1.2485,                0.61603,                0.66624,                0.64983,                0.38158,                 1.4564,                 1.0936,                 1.3251,                   1.51,                 1.1819,                 1.3098,             0.00041839,             0.00041839,             0.00041839
+                      8,                 1.4129,                 1.1561,                 1.2676,                 1.5104,                 1.3392,                 1.2407,                0.61885,                0.66341,                 0.6509,                 0.3745,                 1.5002,                  1.099,                 1.3369,                 1.5551,                 1.1736,                 1.3251,             0.00039087,             0.00039087,             0.00039087
+                      9,                 1.3956,                 1.1182,                 1.2551,                 1.4974,                 1.2879,                 1.2325,                0.65077,                0.69959,                0.68473,                0.41182,                 1.4222,                 1.0129,                 1.2994,                   1.48,                 1.0704,                 1.2851,             0.00036335,             0.00036335,             0.00036335
+                     10,                 1.3789,                 1.0854,                  1.242,                 1.4823,                 1.2438,                 1.2192,                0.64959,                0.70791,                0.69474,                0.42239,                 1.4016,                0.98037,                 1.2826,                 1.4588,                 1.0296,                  1.273,             0.00033582,             0.00033582,             0.00033582
+                     11,                 1.3798,                0.97678,                 1.2634,                 1.4729,                 1.0571,                 1.2425,                0.66872,                0.72864,                0.70739,                  0.435,                  1.381,                0.95833,                 1.2755,                 1.4412,                0.99441,                 1.2626,              0.0003083,              0.0003083,              0.0003083
+                     12,                 1.3564,                0.94245,                 1.2491,                 1.4509,                 1.0138,                 1.2256,                0.66667,                0.74568,                0.72413,                0.44586,                 1.3702,                0.91433,                 1.2578,                  1.441,                0.94525,                 1.2481,             0.00028078,             0.00028078,             0.00028078
+                     13,                 1.3397,                0.91353,                 1.2365,                  1.433,                0.98027,                 1.2159,                0.66883,                0.76455,                0.73376,                0.45962,                  1.349,                0.88796,                 1.2533,                 1.4167,                0.92008,                 1.2452,             0.00025326,             0.00025326,             0.00025326
+                     14,                 1.3208,                0.88717,                 1.2241,                 1.4155,                0.95617,                    1.2,                0.66917,                0.77092,                0.73769,                0.46454,                 1.3385,                   0.87,                 1.2441,                 1.4019,                0.89797,                 1.2303,             0.00022574,             0.00022574,             0.00022574
+                     15,                 1.3062,                0.86752,                 1.2179,                 1.3984,                0.92803,                 1.1968,                0.66989,                0.77004,                0.74171,                0.46884,                 1.3253,                0.85879,                 1.2346,                 1.3883,                0.88954,                 1.2262,             0.00019821,             0.00019821,             0.00019821
+                     16,                 1.2904,                0.84195,                 1.2058,                 1.3888,                0.89869,                 1.1847,                0.68683,                0.77113,                0.74443,                0.47411,                 1.3159,                0.83833,                 1.2319,                 1.3847,                0.86035,                 1.2219,             0.00017069,             0.00017069,             0.00017069
+                     17,                 1.2762,                0.82295,                    1.2,                 1.3719,                0.87611,                 1.1774,                0.69727,                0.77272,                0.74916,                0.47828,                 1.3081,                0.82849,                 1.2272,                  1.373,                0.84983,                 1.2148,             0.00014317,             0.00014317,             0.00014317
+                     18,                 1.2618,                0.80402,                 1.1896,                 1.3544,                 0.8532,                 1.1675,                0.69688,                0.78053,                0.75226,                0.48302,                 1.3011,                0.81901,                 1.2186,                 1.3651,                   0.84,                 1.2065,             0.00011565,             0.00011565,             0.00011565
+                     19,                 1.2466,                 0.7869,                 1.1784,                 1.3404,                 0.8356,                 1.1586,                0.69074,                0.78705,                0.75501,                0.48633,                 1.2951,                0.80962,                 1.2114,                 1.3592,                0.83344,                 1.2037,             8.8126e-05,             8.8126e-05,             8.8126e-05
+                     20,                 1.2327,                0.76884,                 1.1722,                 1.3244,                0.81734,                 1.1528,                0.69436,                 0.7831,                0.75406,                0.48651,                 1.2919,                0.80526,                 1.2104,                 1.3553,                0.82762,                 1.2005,             6.0604e-05,             6.0604e-05,             6.0604e-05

weights/best.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfdf167b393544b058f0d4bc74c772c209c98961e1bf234b4e11c228f50f4269
+size 9260104

weights/best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48301bdc12155643eef9cc044d46457a2441d672d3633d138ce1ffa0267b9253
+size 5764580

weights/last.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d4b7d938465330d47c1fa5b4f6f50b599be613f9c2737a90ed7f93e58629058
+size 5764580