Imageai

This is a major ImageAI release that provides a wider range of APIs, specifically for training and detecting with custom YOLOv3 models on custom datasets.

What's new
- Training of **custom YOLOv3 models** on custom image datasets annotated in Pascal VOC format
- Single and multi-model **mAP** evaluation of saved custom models
- Object detection in images using **custom YOLOv3 models**
- Video Object detection in using **custom YOLOv3 models**
- Video Detection analysis in using **custom YOLOv3 models**
- Support for file and Numpy array inputs/outputs for all custom detection

essential-v4
This release contain supporting files for training custom object detection models and performing detection. The files in this release are:

- **pretrained-yolov3.h5** : A pre-trained YOLOv3 model for transfer learning when training new detection models

- **hololens.zip** : A sample detection dataset of the Hololens with Pascal VOC annotation

- **headset.zip** : A sample detection dataset of Hololens and Oculus headsets with Pascal VOC annotation

- **hololens-ex-60--loss-2.76.h5** : A YOLOv3 model trained with ImageAI on the Hololens dataset

- **detection_config.json** : The configuration JSON file for performing detection in images and video using the trained YOLOv3 model for Hololens.

This is a major ImageAI release that provides a wider range of APIs for custom models and recognition. This wheel is provided for Python 3.x .

What's new:

- Options to set video detection timeout in seconds
- Options to save trained custom model in full
- Support for running custom prediction with fully saved model without specifying network type
- Support for running custom prediction with any fully saved Keras model
- Support for converting custom trained models to Tensorflow (.pb) format
- Support for continuous training from previously saved custom model
- Support for transfer learning from pre-trained models for small datasets
- Only models with increased accuracy will be saved during training
- Support for loading and prediction with multiple custom models
- Support for exporting models to DeepStack format


models-v3
In this release are models used in the sample codes.

This is a major ImageAI release that provides a wider range of APIs . This wheel is provided for Python 3.x .


What's new:
- Bug Fixes

- Option to state image size during custom image prediction model trainings

- Object Detection and Video Object detection now returns bounding box coordinates ('box points') (x1,y1,x2, y2) for each object detected in addition to object's 'name' and 'percentage probability'

- Options to hide 'percentage probability' and/or object 'name' from being shown in detected image or video - Support for video object detection on video live stream from device camera, connected camera and IP camera

- Support for YOLOv3 and TinyYOLOv3 for all object detection and video object detection tasks. - Video object detection for all input types (video file and camera) now allows defining custom functions to execute after each frame, each second and each minute of the video is detected and processed. Also include option to specify custom function at once video is fully detected and processed

- For each custom function specified, ImageAI returns the frame/seconds/minute/full video analysis of the detections that include the objects' details ( name , percentage probability, box_points), number of instance of each unique object detected (counts) and overall average count of the number of instance of each unique object detected in the case of second / minute / full video analysis

- Options to return detected frame at every frame, second or minute processed as a Numpy array.

This is a major <b>ImageAI</b> release that provides a wider range of APIs and the base implementation of future updates. This wheel is provided for Python 3.x . <br>
<br> What's new:
<br>
- Addition of Custom Image Prediction model trainings using SqueezeNet, ResNet50, InceptionV3 and DenseNet121 <br>
- Addition of Custom Image Prediction with custom trained models and generated model class json <br>
- Preview Release: Addition of Video Object Detection and Video Custom Objects Detection (object tracking) <br>
- Addition of file, numpy array and stream input types for all image prediction and object detection tasks (file inputs only for video detections) <br>
- Addition of file and numpy array output types for object detection and custom object detection in images <br>
- Introduction of 4 speed modes ('normal', 'fast', 'faster' and 'fastest') for image prediction, allowing prediction time to reduce by 50% at 'fastest' while maintaining prediction accuracy <br>
- Introduction of 5 speed modes ('normal', 'fast', 'faster', 'fastest' and 'flash') for all object detection and video object detection tasks, allowing detection time to reduce by over 80% <br>
with 'flash' detection accuracy balanced with 'minimum_percentage_probability' which is kept at low values <br>
- Introduction of rate of frame detections, allowing developers to adjust intervals of detections in videos in favour of real time/close to real time result. <br> <br>
<br>

This version of ImageAI supports Image Prediction, Object Detection and Extraction.

This release contains the models pre-trained on IdenProf, the Identifiable Professionals datatset as well as the model JSON class mapping file.