Release/v1.1.0 (#19)

* add signals model
* remove old code & update precommit
* update versions & changelog
* rename model file
* checks on gps and video durations
* fix signal draw in output video
* fix signal draw in output video
* remove python 3.9 support
* update docs
* update manual

Author: durandv

.pre-commit-config.yaml

@@ -3,19 +3,19 @@ default_language_version:
 
 repos:
   - repo: https://github.com/pycqa/isort
-    rev: 5.13.2
+    rev: 6.0.0
     hooks:
       - id: isort
         args: ["--profile", "black", "--filter-files"]
 
   - repo: https://github.com/psf/black
-    rev: 24.10.0
+    rev: 25.1.0
     hooks:
       - id: black
         args: ["--config", ".code_quality/pyproject_black.toml"]
 
   - repo: https://github.com/PyCQA/bandit
-    rev: 1.7.10
+    rev: 1.8.2
     hooks:
       - id: bandit
         args:
@@ -35,11 +35,3 @@ repos:
     hooks:
       - id: prettier
         types: [yaml]
-
-  - repo: https://github.com/PyCQA/docformatter
-    rev: v1.7.5
-    hooks:
-      - id: docformatter
-        additional_dependencies: [tomli]
-        args: [--in-place, --config]
-        exclude: ^(tests|notebooks|docs|models)/

README.md

@@ -1,7 +1,7 @@
 <div align="center">
   <h1>Pavimentados</h1>
 
-[User Manual](docs/manual_202410.pdf)
+[User Manual](docs/manual.pdf)
 
 ![analytics image (flat)](https://raw.githubusercontent.com/vitr/google-analytics-beacon/master/static/badge-flat.gif)
 [![Quality Gate Status](https://sonarcloud.io/api/project_badges/measure?project=EL-BID_pavimentados&metric=alert_status)](https://sonarcloud.io/summary/new_code?id=EL-BID_pavimentados)
@@ -16,7 +16,7 @@ Pavimentados is a tool that allows the identification of pavement faults located
 This library provides an environment to use computer vision models developed to detect different elements. 
 The detected elements are then used to generate metrics that aid the process of planning road maintenance.
 
-The model files can be downloaded from this [link](https://github.com/EL-BID/pavimentados/raw/refs/heads/main/models/model_20240818.tar.gz).
+The model files can be downloaded from this [link](https://github.com/EL-BID/pavimentados/raw/refs/heads/release/v1.1.0/models/model.tar.gz).
 
 > **Important changes**: Unlike the previous version, this new version does not include traffic sign detection. We hope to be able to 
 > include it again in future versions. 
@@ -42,7 +42,7 @@ pip install pavimentados
 ```
 
 Next, 
-* [download the model](https://github.com/EL-BID/pavimentados/raw/refs/heads/main/models/model_20240818.tar.gz) from this link
+* [download the model](https://github.com/EL-BID/pavimentados/raw/refs/heads/release/v1.1.0/models/model.tar.gz) from this link
 
 * Decompress it using the following command
 ```bash
@@ -141,6 +141,7 @@ In the `results` object you will find the following:
  1. table_summary_sections: DataFrame with summary table by sections.
  2. data_resulting: DataFrame with results per frame.
  3. data_resulting_fails: DataFrame with results by unique faults encountered.
+ 4. signals_summary : DataFrame with results by signal.
 
 To see more details about the results please refer to [this page](docs%2Fresults.md).
 

docs/CHANGELOG.md

@@ -5,6 +5,17 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.1.0] - 2025-01-29
+### Added
+- Support for traffic signals detection with a new YoloV11 model.
+- Checks on video and gps durations.
+### Changed
+- Upgraded ultralytics to 8.3.0.
+- Upgraded documentation.
+### Removed
+- Onnx dependencies.
+- Python 3.9 support.
+
 ## [1.0.5] - 2024-12-13
 ### Added
 - Add GPX format in gps sources.

docs/MODELS.md

@@ -20,3 +20,25 @@ The model returns the following information:
  - Probability with which that detection is considered
 
 > **Note**: This model was retrained using the folowing base model: [RoadDamageDetection](https://github.com/oracl4/RoadDamageDetection).
+
+## Detection of signals
+---
+
+This model is based on Yolo version 11, which allows to identify objects or elements within an image. 
+It was retrained from a base model to detect 10 types of signals:
+
+1. PREVENCION
+2. REGLAMENTARIA
+3. INFORMATIVA
+4. CURVA
+5. PERMITE
+6. OBRA
+7. PROHIBICION
+8. ADVERTENCIA
+9. STOP
+10. CEDAPASO
+
+The model returns the following information:
+ - Class of the identified object
+ - Location within the image
+ - Probability with which that detection is considered

docs/manual.pdf

@@ -3,8 +3,8 @@
 
 As a result of processing a video or images, the library returns the following datasets:
 
-
 ## data_resulting
+
 Information about the individual detections made on each image / photogram.
 
 | Field           | Type         | Description                                                                         | Example                                                                        |
@@ -27,9 +27,9 @@ Information about the individual detections made on each image / photogram.
 | total_area      | int          | Total area of the analyzed image.                                                   | 2073600                                                                        |
 | fail_id_section | int          | Identifier of the failure.                                                          | 0                                                                              |
 
-
 ## data_resulting_fails
-Information about each of the failures. A failure is the aggregation of several detections that are considered part of 
+
+Information about each of the failures. A failure is the aggregation of several detections that are considered part of
 the same problem (failures have a single type and can have many detections).
 
 | Field            | Type   | Description                                                               | Example                                     |
@@ -47,8 +47,8 @@ the same problem (failures have a single type and can have many detections).
 | width            | float  | Average width of faults measured in pixels.                               | 789.3660466811236                           |
 | boxes            | int    | Number of bounding boxes representing the failure.                        | 19                                          |
 
-
 ## table_summary_sections
+
 Information about what is detected every 100 meters.
 
 | Field            | Type  | Description                                        | Example Value       |
@@ -65,3 +65,18 @@ Information about what is detected every 100 meters.
 | end_longitude    | float | Ending longitude of the section.                   | -77.15626066666667  |
 | end_latitude     | float | Ending latitude of the section.                    | -11.7830935         |
 | section_distance | float | Distance of the section in meters.                 | 106.6830640804603   |
+
+## signals_summary
+
+Information about signals detected in each frame.
+
+| Field              | Type   | Description                                              | Example                      |
+|--------------------|--------|----------------------------------------------------------|------------------------------|
+| fotogram           | int    | Frame number of the video.                               | 52                           |
+| position_boxes     | string | Position of the bounding box in the frame.               | left                         |
+| score              | float  | Confidence score of the detection.                       | 0.40509921312332153          |
+| latitude           | float  | Latitude of the location.                                | 10.380599907222223           |
+| longitude          | float  | Longitude of the location.                               | -84.36906742202616           |
+| boxes              | string | Bounding box coordinates in the format [x1, y1, x2, y2]. | [0.540, 0.205, 0.550, 0.219] |
+| signal_class_names | string | Name of the signal class.                                | ADVERTENCIA                  |
+| ID                 | int    | Unique identifier of the detection.                      | 0                            |

docs/manual_202410.pdf

@@ -3,5 +3,10 @@
 		"yolo_threshold": 0.20,
 		"yolo_iou": 0.45,
 		"yolo_max_detections": 100
+	},
+	"signal_model": {
+		"yolo_threshold": 0.40,
+		"yolo_iou": 0.45,
+		"yolo_max_detections": 100
 	}
 }

docs/results.md

@@ -13,7 +13,7 @@
     stack_columns_dataset,
 )
 
-first_aggregation_dict = {"ind": "count"}  # , "perc_area": "sum"
+first_aggregation_dict = {"ind": "count"}
 
 second_aggregation_dict = {
     "distances": "sum",
@@ -162,31 +162,8 @@ def generate_final_results_signal(
         gps_obj,
         classes_names_yolo_signal=None,
     ):
-        # Reemplazo de clases
-        # signals_class_mapping = config["signals_class_mapping"]
-
-        signal_base_predictions = results_obj["signal_base_predictions"]
-        final_signal_classes = results_obj["final_signal_classes"]
-        state_predictions = results_obj["state_predictions"]
-
-        # signal_base_predictions = [
-        #     [signals_class_mapping[item] if item in signals_class_mapping.keys() else item for item in sublist]
-        #     for sublist in results_obj["signal_base_predictions"]
-        # ]
-        # final_signal_classes = [
-        #     [signals_class_mapping[item] if item in signals_class_mapping.keys() else item for item in sublist]
-        #     for sublist in results_obj["final_signal_classes"]
-        # ]
-        # state_predictions = [
-        #     [signals_class_mapping[item] if item in signals_class_mapping.keys() else item for item in sublist]
-        #     for sublist in results_obj["state_predictions"]
-        # ]
-
         BOXES_SIGNAL = results_obj["boxes_signal"]
         CLASSES_SIGNAL = results_obj["classes_signal"]
-        SIGNAL_CLASSES_siames = final_signal_classes
-        SIGNAL_CLASSES_BASE = signal_base_predictions
-        SIGNAL_STATE = state_predictions
         SCORES_SIGNAL = results_obj["scores_signal"]
         final_latitude = gps_obj.gps_df.latitude.values
         final_longitude = gps_obj.gps_df.longitude.values
@@ -195,11 +172,11 @@ def generate_final_results_signal(
         # Nos dice el cuadrante.
         def position(center):
             if center < 0.33:
-                return 0
+                return "left"
             elif center < 0.66:
-                return 1
+                return "center"
             else:
-                return 2
+                return "right"
 
         position_boxes = [[position((box[0] + box[2]) / 2) for box in BOXES_SIGNAL[f]] for f in range(len(BOXES_SIGNAL))]
 
@@ -210,9 +187,6 @@ def position(center):
                     f,
                     position_boxes[f][i],
                     SCORES_SIGNAL[f][i],
-                    SIGNAL_STATE[f][i],
-                    SIGNAL_CLASSES_siames[f][i],
-                    SIGNAL_CLASSES_BASE[f][i],
                     int(CLASSES_SIGNAL[f][i]),
                     final_latitude[f],
                     final_longitude[f],
@@ -226,34 +200,19 @@ def position(center):
                 "fotogram",
                 "position_boxes",
                 "score",
-                "signal_state",
-                "signal_class_siames",
-                "signal_class_base",
                 "signal_class",
                 "latitude",
                 "longitude",
                 "boxes",
             ],
         )
 
-        df["signal_class_siames_names"] = df["signal_class_siames"].values
         df["signal_class_names"] = df["signal_class"].apply(lambda x: classes_names_yolo_signal[x])
-
-        yolo_classes_to_keep = []
-
-        x = tuple(zip(df["signal_class_siames_names"].values, df["signal_class_names"].values))
-
-        final_classes = []
-        for i in range(len(x)):
-            if x[i][1] in yolo_classes_to_keep:
-                final_classes.append(x[i][1])
-            else:
-                final_classes.append(x[i][0])
-
-        df["final_classes"] = final_classes
+        df["signal_class_names"] = df["signal_class_names"].apply(lambda x: config["signals_class_mapping"][x])
+        df.drop(columns=["signal_class"], inplace=True)
         df["ID"] = range(len(df))
 
-        df = df.sort_values(["final_classes", "position_boxes", "fotogram"]).reset_index(drop=True)
+        df = df.sort_values(["signal_class_names", "position_boxes", "fotogram"]).reset_index(drop=True)
 
         # cantidad de fotogramas a sacar repetidas.
         N_fotogram = 5
@@ -262,7 +221,7 @@ def position(center):
 
         for i in range(len(df) - 1, 0, -1):
             if (
-                (df.loc[i - 1, "final_classes"] == df.loc[i, "final_classes"])
+                (df.loc[i - 1, "signal_class_names"] == df.loc[i, "signal_class_names"])
                 & (df.loc[i - 1, "position_boxes"] == df.loc[i, "position_boxes"])
                 & (
                     (np.abs(df.loc[i - 1, "fotogram"] - df.loc[i, "fotogram"]) <= N_fotogram)
@@ -271,7 +230,7 @@ def position(center):
                         <= meters_dist
                     )
                 )
-                & (df.loc[i, "final_classes"] != "OTRO")
+                & (df.loc[i, "signal_class_names"] != "OTRO")
             ):
                 df.loc[i - 1, "ID"] = df.loc[i, "ID"]
 

models/model.tar.gz

@@ -23,6 +23,7 @@ def __init__(self):
 
     def _calculate_seconds_from_start(self):
         self.gps_df["seconds_from_start"] = self.gps_df.seconds.values - self.gps_df.seconds.values[0]
+        self.gps_duration = self.gps_df["seconds_from_start"].max()
 
     def adjust_gps_data(self, number_images):
         # Interpolación lineal por cada segundo (corrección de outliers)
@@ -230,7 +231,7 @@ def get_gpsdata_from_exif(self, filename):
                 try:
                     data = exif[key].decode() if isinstance(exif[key], bytes) else exif[key]
                     exif_data[name] = data
-                except:
+                except:  # noqa: E722
                     logger.debug(f"Error decoding exif in key: {key}")
 
             if "GPSInfo" in exif_data:
@@ -258,7 +259,7 @@ def load_single_value(self, img_path):
                 lat = decimal_coords(d["GPSLatitude"], d["GPSLatitudeRef"])
                 lon = decimal_coords(d["GPSLongitude"], d["GPSLongitudeRef"])
                 time = dt.datetime.strptime(d["DateTimeOriginal"], "%Y:%m:%d %H:%M:%S")
-            except:
+            except:  # noqa: E722
                 raise InvalidGPSData(f"Could not load GPS data from image: {img_path}")
 
         return {"timestamp": time, "longitude": lon, "latitude": lat}

models/model_20240818.tar.gz

@@ -26,20 +26,12 @@
 	},
 
 	"signal_model": {
-		"enabled": false,
-		"path": "signal_model/yolov8-signals-240707-2106",
+		"enabled": true,
+		"path": "signal_model/yolo11-signals-datasetWOotherysem-250115-2342",
 		"model_filename": "model.pt",
 		"classes_filename": "classes.names",
 		"yolo_threshold": 0.10,
 		"yolo_iou": 0.45,
 		"yolo_max_detections": 100
-	},
-
-	"siamese_model": {
-		"enabled": false,
-		"path": "siamese_model/siamese_128",
-		"model_filename": "onnx_siamese_model.onnx",
-		"embeddings_filename": "embeddings_references.pickle",
-		"image_size": [128, 128, 3]
 	}
 }