def run():
    # Determine path to configuration file.
    local_dir = os.path.dirname(__file__)
    config_path = os.path.join(local_dir, 'novelty_config')
    # Note that we provide the custom stagnation class to the Config constructor.
    config = neatfast.Config(neatfast.DefaultGenome,
                             neatfast.DefaultReproduction,
                             neatfast.DefaultSpeciesSet,
                             neatfast.DefaultStagnation, config_path)

    ne = NoveltyEvaluator(4, 'color')
    if ne.scheme == 'color':
        config.output_nodes = 3
    else:
        config.output_nodes = 1

    pop = neatfast.Population(config)

    # Add a stdout reporter to show progress in the terminal.
    pop.add_reporter(neatfast.StdOutReporter(True))
    stats = neatfast.StatisticsReporter()
    pop.add_reporter(stats)

    while 1:
        pop.run(ne.evaluate, 1)

        winner = stats.best_genome()
        if ne.scheme == 'gray':
            image = eval_gray_image(winner, config, full_scale * width,
                                    full_scale * height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, config, full_scale * width,
                                     full_scale * height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, config, full_scale * width,
                                    full_scale * height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        im = np.clip(np.array(image), 0, 255).astype(np.uint8)
        im = ne.image_from_array(im)
        im.save('winning-novelty-{0:06d}.png'.format(pop.generation))

        if ne.scheme == 'gray':
            image = eval_gray_image(winner, config, width, height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, config, width, height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, config, width, height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        float_image = np.array(image, dtype=np.float32) / 255.0
        ne.archive.append(float_image)
    def make_high_resolution(self, genome, config):
        genome_id, genome = genome

        # Make sure the output directory exists.
        if not os.path.isdir('rendered'):
            os.mkdir('rendered')

        if self.scheme == 'gray':
            image_data = eval_gray_image(genome, config, self.full_width,
                                         self.full_height)
        elif self.scheme == 'color':
            image_data = eval_color_image(genome, config, self.full_width,
                                          self.full_height)
        else:
            image_data = eval_mono_image(genome, config, self.full_width,
                                         self.full_height)

        image = self.make_image_from_data(image_data)
        pygame.image.save(
            image,
            "rendered/rendered-{}-{}.png".format(os.getpid(), genome_id))

        with open("rendered/genome-{}-{}.bin".format(os.getpid(), genome_id),
                  "wb") as f:
            pickle.dump(genome, f, 2)
Beispiel #3
0
def evaluate_lowres(genome, config, scheme):
    if scheme == 'gray':
        return eval_gray_image(genome, config, width, height)
    elif scheme == 'color':
        return eval_color_image(genome, config, width, height)
    elif scheme == 'mono':
        return eval_mono_image(genome, config, width, height)

    raise Exception('Unexpected scheme: {0!r}'.format(scheme))
def evaluate(genome, scheme):
    if scheme == 'gray':
        return eval_gray_image(genome, width, height)
    elif scheme == 'color':
        return eval_color_image(genome, width, height)
    elif scheme == 'mono':
        return eval_mono_image(genome, width, height)

    raise Exception('Unexpected scheme: {0!r}'.format(scheme))
Beispiel #5
0
def run():
    cfg = config.Config('novelty_config')
    cfg.pop_size = 100
    cfg.max_fitness_threshold = 1e38

    ne = NoveltyEvaluator(4, 'gray')
    if ne.scheme == 'color':
        cfg.output_nodes = 3
    else:
        cfg.output_nodes = 1

    pop = population.Population(cfg)

    while 1:
        pop.run(ne.evaluate, 1)

        winner = pop.statistics.best_genome()
        if ne.scheme == 'gray':
            image = eval_gray_image(winner, full_scale * width,
                                    full_scale * height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, full_scale * width,
                                     full_scale * height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, full_scale * width,
                                    full_scale * height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        im = np.clip(np.array(image), 0, 255).astype(np.uint8)
        im = ne.image_from_array(im)
        im.save('winning-novelty-{0:06d}.png'.format(pop.generation))

        if ne.scheme == 'gray':
            image = eval_gray_image(winner, width, height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, width, height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, width, height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        float_image = np.array(image, dtype=np.float32) / 255.0
        ne.archive.append(float_image)
def run():
    cfg = config.Config('config')
    cfg.pop_size = 100
    cfg.max_fitness_threshold = 1e38

    ne = NoveltyEvaluator(4, 'gray')
    if ne.scheme == 'color':
        cfg.output_nodes = 3
    else:
        cfg.output_nodes = 1

    pop = population.Population(cfg)

    while 1:
        pop.run(ne.evaluate, 1)

        winner = pop.statistics.best_genome()
        if ne.scheme == 'gray':
            image = eval_gray_image(winner, full_scale * width, full_scale * height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, full_scale * width, full_scale * height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, full_scale * width, full_scale * height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        im = np.clip(np.array(image), 0, 255).astype(np.uint8)
        im = ne.image_from_array(im)
        im.save('winning-novelty-{0:06d}.png'.format(pop.generation))

        if ne.scheme == 'gray':
            image = eval_gray_image(winner, width, height)
        elif ne.scheme == 'color':
            image = eval_color_image(winner, width, height)
        elif ne.scheme == 'mono':
            image = eval_mono_image(winner, width, height)
        else:
            raise Exception('Unexpected scheme: {0!r}'.format(ne.scheme))

        float_image = np.array(image, dtype=np.float32) / 255.0
        ne.archive.append(float_image)
    def make_high_resolution(self, genome):
        # Make sure the output directory exists.
        if not os.path.isdir('rendered'):
            os.mkdir('rendered')

        if self.scheme == 'gray':
            image_data = eval_gray_image(genome, self.full_width, self.full_height)
        elif self.scheme == 'color':
            image_data = eval_color_image(genome, self.full_width, self.full_height)
        else:
            image_data = eval_mono_image(genome, self.full_width, self.full_height)

        image = self.make_image_from_data(image_data)
        pygame.image.save(image, "rendered/rendered-{}-{}.png".format(os.getpid(), genome.ID))

        with open("rendered/genome-{}-{}.bin".format(os.getpid(), genome.ID), "wb") as f:
            pickle.dump(genome, f, 2)
Beispiel #8
0
    def evaluate(self, genomes, config):
        jobs = []
        for genome_id, genome in genomes:
            jobs.append(
                self.pool.apply_async(evaluate_lowres,
                                      (genome, config, self.scheme)))

        new_archive_entries = []
        for (genome_id, genome), j in zip(genomes, jobs):
            image = np.clip(np.array(j.get()), 0, 255).astype(np.uint8)
            float_image = image.astype(np.float32) / 255.0

            genome.fitness = (width * height)**0.5
            for a in self.archive:
                adist = np.linalg.norm(float_image.ravel() - a.ravel())
                genome.fitness = min(genome.fitness, adist)

            if random.random() < 0.02:
                new_archive_entries.append(float_image)
                #im = self.image_from_array(image)
                #im.save("novelty-{0:06d}.png".format(self.out_index))

                if self.scheme == 'gray':
                    image = eval_gray_image(genome, config, full_scale * width,
                                            full_scale * height)
                elif self.scheme == 'color':
                    image = eval_color_image(genome, config,
                                             full_scale * width,
                                             full_scale * height)
                elif self.scheme == 'mono':
                    image = eval_mono_image(genome, config, full_scale * width,
                                            full_scale * height)
                else:
                    raise Exception('Unexpected scheme: {0!r}'.format(
                        self.scheme))

                im = np.clip(np.array(image), 0, 255).astype(np.uint8)
                im = self.image_from_array(im)
                im.save('novelty-{0:06d}.png'.format(self.out_index))

                self.out_index += 1

        self.archive.extend(new_archive_entries)
        print('{0} archive entries'.format(len(self.archive)))
    def evaluate(self, genomes):
        jobs = []
        for genome in genomes:
            jobs.append(self.pool.apply_async(evaluate, (genome, self.scheme)))

        new_archive_entries = []
        for g, j in zip(genomes, jobs):
            image = np.clip(np.array(j.get()), 0, 255).astype(np.uint8)
            float_image = image.astype(np.float32) / 255.0

            g.fitness = (width * height) ** 0.5
            for a in self.archive:
                adist = np.linalg.norm(float_image.ravel() - a.ravel())
                g.fitness = min(g.fitness, adist)

            if random.random() < 0.02:
                new_archive_entries.append(float_image)
                #im = self.image_from_array(image)
                #im.save("novelty-{0:06d}.png".format(self.out_index))

                if self.scheme == 'gray':
                    image = eval_gray_image(g, full_scale * width, full_scale * height)
                elif self.scheme == 'color':
                    image = eval_color_image(g, full_scale * width, full_scale * height)
                elif self.scheme == 'mono':
                    image = eval_mono_image(g, full_scale * width, full_scale * height)
                else:
                    raise Exception('Unexpected scheme: {0!r}'.format(self.scheme))

                im = np.clip(np.array(image), 0, 255).astype(np.uint8)
                im = self.image_from_array(im)
                im.save('novelty-{0:06d}.png'.format(self.out_index))

                self.out_index += 1

        self.archive.extend(new_archive_entries)
        print('{0} archive entries'.format(len(self.archive)))